ABSTRACT

Background

Clinical and structural correlates of lateralized interictal delta activity in patients with temporal lobe epilepsy (TLE) have been well documented in the literature. Nevertheless, its occurrence has not been considered a significant clinical feature.

Objective

To evaluate the significance of focal delta- range slowing for localizing the epileptogenic focus in patients with TLE, and predicting the outcome of temporal lobe surgery.

Methods

Sixteen consecutive patients with temporal lobe epilepsy who underwent anterior temporal lobe resections were selected for the study. Findings of MRI, SPECT, neuropyschology assessment, pathology and surgical outcome were analyzed and correlated with focal delta slow activity of background rhythm.

Results

Twelve of 16 patients (75%) had localized temporal delta slowing corresponding to the resection site and pathology. Temporal delta slowing was the most frequent interictal EEG finding (75%) compared to spike and sharp wave (44%). MRI showed concordant abnormalities in 75% of the patients, and neuropyschology testing was able to lateralize the involved hemisphere in 37.5%. SPECT was concordant in 56%. There was no false localization with temporal delta activity. Slow wave EEG had a higher marginal probability than neuropsychological assessment of predicting the focus, and was equally effective as other investigative methods.

Conclusion

These results suggest that focal temporal delta slowing is useful in the localization of epileptogenic foci. There was no discordance with the resection site and pathology.

RESUME

Introduction

Les corrélations entre clinique, et topographie des ondes delta intercritiques au cours de l’épilepsie temporal (TLE) ont été bien documentées dans la litterature.

But

Évaluer la relation entre ralentissement delta et la localisation d’un foyer épileptogène dans le cadre d’une TLE et après une lobectomie temporale.

Méthodes

Seize patients consécutifs atteints d’une épilepsie temporale qui ont bénéficié d’une lobectomie temporale ont été séletionnés. Les données IRM, SPECT, neuropsychologiques, neuropathologiques et l’évolution post-chirurgicale ont été analysées et correlées au foyer d’ondes delta.

Résultats

Douze des 16 patients (75%) avaient une correspondance entre les ondes delta, et la résection chirurgicale et l’anatomopathologie. Les ondes lentes delta temporales sont les anomalies les plus fréquemment rencontrées durant les périodes intercriques comparées aux spikes et pointes ondes (44%). La concordance entre l’IRM et les anomalies est notée chez 75% des patients ; les tests neuropsychologiques sont capables de déterminer la lateralisation de l’hémisphère impliqué dans 37% des cas. La SPECT était concordante dans 56% des cas. Il n’y avait pas de faux positif dans la localisation par l’activité temporale delta. Les ondes EEG lentes ont une plus grande probabilité localisatrice que les tests neuropsychologiques et sont aussi effectives que les autres méthodes

Conclusion

Ces résultats suggèrent que les ondes lentes temporales sont utiles dans la localization d’un foyer épileptique. Il n’y a aucune discordance entre le site de la résection et l’anatomo-pathologie.

Keywords : EEG, Epilepsie, Crises épileptiques, Chirurgie. Lobe temporal, slowing, EEG, Epilepsy, Seizure, Surgery,Temporal lobe

INTRODUCTION

Resective surgery has proved to be an effective therapy for relief of intractable TLE [6,19,27,36]. Success in epilepsy surgery depends in large on the accuracy of pre-operative localization of the epileptogenic focus prior to cortical excision. This is primarily achieved by examining repeated preoperative electroencephalograph (EEG) studies, and additional independent data, which include clinical features of attack, evidence of localized brain dysfunction in neuropsychological assessment, regional hypo- perfusion demonstrated by SPECT, and structural lesions by MRI [17,28,45,46]. Congruence of data from all available parameters is crucial for seizure free outcome postoperatively.

Although irritative activity in an EEG is still considered to be a corner stone for localization of an epileptogenic focus, it is often the case that interictal spikes as well as the electrographic onset of seizures are widespread or bilateral [6,29,37,38,42].

The literature places little emphasis on the localizing value of EEG background rhythm disturbance. Engel and colleagues [16,18] describe several EEG criteria such as interictal spikes, thiopental activation of spike and fast activity and after-discharge-threshold determination, excluding background analysis. Focal delta slow waves on scalp EEG, either continuous and polymorphic or intermittent and rhythmic, are generally considered to be an indication of localized structural lesion involving cortical gray matter [24,47] or subcortical white matter [21]. Moreover, background abnormalities are occasionally observed in epileptic patients and considered evidence of nonspecific functional deficit resulting from underlying cerebral pathology [34,35]. In temporal lobe epilepsy without mass lesion interictal focal delta slowing was not related to the severity of epilepsy or to the neuronal loss of mesial temporal lobe structures. Koutrumanides [30] found a strong correlation between focal delta slowing and temporal lobe hypometabolism on FDG PET, suggesting a focal area of neuronal inhibition that can participate in interictal and ictal activity.

We wished to determine the significance of interictal focal temporal slowing in 16 consecutive patients who underwent diagnostic evaluation between 1993-1996 (University of Miami Epilepsy Program) to localize a resectable epileptic focus. All patients had a surgical resection of temporal epileptogenic cortical lesion. All patients had a regular follow up in the outpatient’s clinic as part of the postoperative protocol. The value of interictal temporal slow delta activity on a surface EEG was examined and correlated with the results of other diagnostic procedures.

PATIENTS AND METHODS (p)

A cohort of 16 consecutive adult patients with TLE (9 males, ages between 23 and 49 years, and 7 females between 27 and 51 years), and a minimal follow up of 2 years after temporal lobectomy were included. All patients had detailed inpatient video-EEG monitoring, MRI of the brain [12,13,26,33], SPECT (both postictal and interictal were reported as a single finding) [9,25,40,41,44], and neuropsychological assessment [11,23,32,43] before temporal lobectomy. Clinical informations obtained from the data base included patient age, sex, side of the epilepsy surgery, and pathologic diagnosis of resected tissue [8,15,31]. Treatment with anticonvulsants was continued postoperatively. Patient follow-up occurred at three-month intervals and included physical examinations and scalp EEG recordings. Surgical outcome was assessed based on Engel’s classification, with class 1 representing freedom from disabling seizure (excellent results) and class 4 no worthwhile improvement (poor results).

Electrophysiological studies

All patients were monitored continuously with 24 hours EEG-Video monitoring using 64 channels Nicolet BMSI (Madison, WI) telemetry system. Scalp electrodes were applied according to international 10-20 System of placement. The patients had at least three seizures with clinical and electrographic evidence. All the EEG data were reviewed in our Neurophysiology laboratory (University of Miami). EEG analyses of awake and sleep record were based on visual observation to quantify periods of paroxysmal activities. The following elements were evaluated:

1. Ictal activity, defined as sustained rhythmic change in EEG activity preceding or coinciding with clinical seizure manifestation.
   
2. Interictal activity
   
   a.Spikes and sharp waves
   
   b.Brief periods of focal delta slowing not contaminated with postictal slowing (> 8 hrs. post seizure onset), spikes or artifact.

Multiple occurrences of focal delta slowing of background rhythm lasting more than 5 seconds were considered significant for analysis. Records were considered positive if more than five epochs of focal temporal delta slow activity were noted.
Statistical Analysis

For each patient in the study, two binary variables (detect/no detect; cure/no cure) were recorded for each of the diagnostic methods (slow wave EEG, MRI, SPECT, Neuropsychological Testing). For patients in whom lesion removal did not result in a cure (i.e., Engel class 1 outcome), the outcome was recorded as a “no cure”.
Since several diagnostic methods were recorded for each patient, repeated measures were utilized when comparing the effectiveness of diagnostic tests. Fisher’s exact test, evaluating marginal homogeneity of dependent data, was conducted to test whether diagnostic methods were equally effective in predicting a cure. Additionally, simple diagnostics, such as sensitivity and specificity, and logit models with diagnostic method as independent variable were employed to study the relationship between the response variable and each of the diagnostic tests.

RESULTS

All sixteen patients underwent surgery. Nine patients had a right and seven had a left anterior temporal lobectomy. All patients had postoperative follow up ranging between two to five years. Twelve patients were seizure free more than 2 years following surgery. Nine out of 12 patients (75%) had focal temporal delta slowing concordant with the resection site, indicating a high predictive value of focal temporal slowing for good surgical outcome (class 1).

Electrographic Data. A scalp EEG was analyzed for both interictal and ictal changes at the time of onset of clinical seizure during EEG video monitoring, and for a buildup of sustained rhythmic discharges. Localized ictal and interictal EEG changes (i.e., focal temporal spikes) were observed in seven of the 16 patients (44%). All were concordant with the resection site and pathology results. Four patients had bilateral electrographic onset, and for three of these 4 patients intracranial EEG recordings were obtained to determine a single focus.

Focal temporal delta slowing, ipsilateral to the resection site, was noted in twelve patients (75%). Nine of the 12 patients with temporal delta slowing had class 1 outcome (75%). It is important to note that only seven of them had combined focal delta slowing and interictal spikes concordant with resection site.

MRI. Evidence of unilateral temporal lobe pathology, ipsilateral to the resection site, was noted on the brain MRI of twelve patients (75%) in the form of increased hippocampal signal on T2 weighted images or hippocampal formation atrophy. Nine of them showed focal delta slowing. Six out of these 9 (66%) were seizure free. MRI was normal in 4 patients for whom scalp EEG failed to provide lateralizing information. In these 4 patients, pathology reported mild gliosis and few ectopic neuron (n = 2), no specific changes (n = 1), and mesial temporal sclerosis (n = 1). Three of the 4 patients with normal MRI exhibited focal delta slowing concordant with resection site (75%) and were seizure free.

SPECT. Qualitative visual analysis of SPECT images disclosed unilateral regional activity concordant with the resection site in nine patients (56%). Three patients had contralateral results and 4 reported as normal. Six out of 9 patients (66%) with positive SPECT had concordant focal temporal delta slowing.

Neuropsychological Tests. Detailed neuropsychological testing was conducted in all patients. For the purpose of our current analysis, only final conclusions of the neuropychology assessment in regard to lateralization were considered. Neuropsychology tests correctly lateralized the epileptogenic focus in six patients (37.5%). For three of 16 patients (19%) neuropsychological testing failed to report lateralization, even though two patients had a clear seizure focus, one in the left temporal lobe and one in the right. False lateralization was obtained for one patient with bilateral onset and diffuse slowing.

In addition, WADA tests were performed a) to evaluate hemisphere dominance in 3 left-handed patients; b) because 2 patients had bilateral ictal onset; c) because 2 patients had diffuse cognition impairment during neuropsychological testing; and d) because 2 patients had verbal and language deficits accompanied by seizure onset in the left temporal lobe.

Pathology. Brain tissue samples of sixteen patients were available for review. Eight specimens (50%) revealed mesial temporal sclerosis (MTS) on routine pathological examination. Six of the 8 patients (75%) exhibited focal delta slowing in EEG recordings. In four specimens few ectopic neuron, minimal gliosis, and old scar with gliosis (a cortical cicatrix) were noted. Three of the 4 patients with gliosis exhibited concordant focal delta slowing. One specimen showed a dysembryoplastic neuroepithelial tumor. In the three remaining specimens no significant changes were observed. However, two of these 3 patients had focal delta slowing. Comparing the brain tissue pathology with the possible risk factors identified in the patients’ history, four cases of MTS were associated with head trauma and three cases with febrile illness; one case was cryptogenic in etiology. Cortical cicatrix was associated with herpes encephalitis. In two cases of cryptogenic etiology and in one case with head trauma no specific changes were observed in the brain tissue pathology. Two cases of minimal gliosis and few ectopic neurons were associated with cryptogenic etiology.

Statistical analysis (p)
The first goal of this study was to determine whether a Slow Wave EEG had the same probability of predicting a positive outcome as the other diagnostic tests. The marginal probability of a positive outcome based on Slow Wave EEG was compared with the marginal probability of a positive outcome based on each diagnostic test separately. Because of the within-subject design, explicit account was taken of the dependence between the variables.

Slow Wave EEG has a higher marginal probability than neuropsychological evaluation in predicting a cure (p = 0.1) and is otherwise equally effective as other methods. It is important to note that the probability value for slow wave EEG versus MRI is 1.0. This suggests that Slow Wave EEG and MRI are equally effective (table 1). In general, there is strong agreement between the two tests and when discordant diagnoses occur the differences are due to random variation.

Sensitivity and Specificity. In general, sensitivity was relatively high for all diagnostic tests. As shown in table 2, slow wave EEG has the highest sensitivity and specificity of any of the diagnostic tools. Sensitivity would be even higher if we excluded cases in which removal of a lesion did not result in a cure. The poor specificity of the diagnostic tests was due to instances for which a lesion removal (detected by other means) resulted in a cure. As an example, all four patients for whom MRI failed to detect a lesion were cured because the lesion was detected by other diagnostic tests. By way of comparison, pathology had a sensitivity of .77 and a specificity of .33.

Logit Model. The second goal was to investigate how likely it was for a cure (seizure freedom) to occur when a lesion was detected as compared to undetected by a single diagnostic measure. The Logit Model analysis suggested that none of the diagnostic tests were effective in predicting a cure (table 3). For example, a cure was equally likely to occur when a lesion was detected (75%) or undetected (75%) by the slow wave EEG recording. The Logit Model analysis is hampered by the small number of negative outcomes reported for the diagnostic tests reflected in the poor specificity of the tests. Note that the MRI had no true negatives and hence a Logit Model analysis was not possible for this diagnostic tool.

DISCUSSION

Accurate localization of the epileptogenic focus is essential prior to surgical treatment for intractable epilepsy [6,17,19,27,28,36,45,46]. In our study, focal temporal delta slowing has been shown to be of great value for seizure localization (75%). This finding of temporal delta slowing of background rhythm is highly correlated with MRI (75%) and SPECT (56%). Also, our study has shown a good surgical outcome (i.e., seizure free) in patients with ipsilateral temporal delta slowing and ipsilateral spikes and sharp waves in their EEG recordings. This indicates that, in patients with temporal lobe epilepsy (TLE), the evaluation of background abnormalities is valuable in localizing the epileptogenic focus.

The neurophysiological mechanisms responsible for the generation of focal delta slowing in TLE are not clear. The observation of a coincidence between spikes and intermittent delta activity suggests that the region of the brain responsible for epileptic discharges may also produce pathologic slow waves in the scalp EEG. Ajmone- Marsan [4] emphasized that, in epileptic patients, nonspecific slow transients or even subcontinuous slow waves may often be the expression of functional changes reflecting an epileptogenic process. Further evidence of this hypothesis was provided by previous studies with simultaneous scalp and an intracranial recording (extradural, sudural and depth electrodes) [1,2,3,5,7]. On the basis of these simultaneous scalp and depth recordings, epileptiform patterns recorded by intracranial electrodes appeared less likely to occur in the surface reflection of the scalp EEG recording. However, some of the scalp recordings appeared as typical epileptiform discharge, mainly in the form of slow transient. Specific factors determining whether an epileptiform discharge will appear at the scalp with typical morphology or in the form of slow transients have not been identified.

The stability of polymorphic delta activity has been investigated utilizing intracerebral electrodes. No significant day-to-day changes in background activity have been shown even though spike and seizure frequency fluctuated [22]. Dally [14] showed that, in general, sleep does not affect the stability of polymorphic delta activity. These results emphasize the stability of background activity in contrast to the unpredictable occurrence of spikes and seizure, enhancing the value of focal delta slowing of background rhythm as a diagnostic tool.

Reiher et al [39] reported that temporal intermittent rhythmic delta (TIRDA) is an accurate indicator of partial seizures. Their study sample consisted of 115 consecutive patients with a clinical diagnosis of complex partial seizures, without identification of the location of the epileptogenic zone. No patients exhibited evidence of clinical seizures activity during run of TIRDA and the TIRDA generally occurred ipsilateral to unilateral temporal spikes. Some of their patients had extratemporal lobe epileptiform discharge. Over all our results support their conclusion which suggested that TIRDA is an indication of temporal lobe seizure. Although our series included morphologies of delta activity other than the characteristic TIRDA, on many occasions focal delta slowing was indeed TIRDA.

Our results were in overall agreement with Gambardella et al [20], who reported that trains of delta waves over temporal regions were observed in >90% of patients with mesiotemporal atrophy assessed by volumetric MRI. In contrast, Bernasconi et al [10] correlated volumetric MRI and MRS of the temporal lobe with interictal focal delta activity in 34 TLE patients and 10 controls. They reported no significant difference in the amount of delta activity in temporal lobe between the controls and patients, nor did they find a correlation between delta activity and neuroimaging measures. Delta activity was not explained by reduced volume of temporal lobe white matter, gray matter, or by abnormalities seen in MRS. Although our study did not include volumetric MRI or MRS, a striking concordance was observed between focal delta activity and the epileptogenic focus. Several of our patients with a normal MRI exhibited focal delta activity, supporting the notion that focal temporal delta activity should be conceptualized as a distinct electrographic phenomenon, directly related to epileptogenic abnormality.

With regard to the follow up of our patients, twelve were seizure free (class 1) more than 2 years following surgery, and nine of them (75%) showed presurgical temporal delta slowing concordant with resection site. In seven patients (58%) ipsilateral spikes/sharp waves were observed in addition to the focal delta slowing. In conclusion, our results strongly suggest that focal delta slowing of background rhythm over the temporal lobe is a reliable and accurate indicator of an epileptogenic focus and ranks equal in accuracy with other standard diagnostic tests and therefore should be included among EEG criteria during standard preoperative evaluation of epilepsy patients.

Table 1. Fisher s Exact Test Comparing Marginal Probability of Detecting a Lesion for Slow Wave EEG versus other Diagnostic Methods

Table 2. Sensitivity (Probability of Positive Outcome Given Lesion Detection) and Specificity (Probability of Negative Outcome Given No Lesion Detection) for Diagnostic Tests

Table 3. Logistic Analysis of Probability of a cure. The independent variable is the binary outcome (detect/no detect) of the diagnostic test

Table 4. Slow wave EEG by outcome