ELECTROCLINICAL CHARACTERISTICS OF MRI NEGATIVE FOCAL EPILEPSY: A VIDEO-EEG STUDY ELEKTROKLINIČKE KARAKTERISTIKE MRI NEGATIVNIH FOKALNIH EPILEPSIJA: VIDEO-EEG MONITORING STUDIJA

Background/Aim. Epileptogenic lesions carry intrinsic epileptogenicity or epileptogenic potential in their close vicinity. One third of patients with focal epilepsy have no epileptogenic lesions magnetic resonance imaging [MRI(-)]. The aim of this study was to determine the epileptogenic zone investigating electrical and clinical properties of MRI- patients. Methods. In 180 patients with focal epilepsy we analyzed 1,712 seizures for interictal and ictal electroencephalography (EEG) and seizure semiology. If multiple seizures occurred we took the best seen on video as an example, with secondary generalized tonic-clonic seizures (GTCS) if it occurred. Brain MRI was focused to investigate the zone of ictal EEG onset. Electroclinical properties of the MRIpatients were compared to lesion positive patients [MRI(+)]. Results. A single epileptogenic lesion was identified in 68.89% [hippocampal sclerosis (HS) in 58, focal cortical dysplasia (FCD) in 28 and other pathologies in 38 patients]. MRI(-) patients had significantly less interictal epileptiform abnormalities, and presented more often (p < 0.001) with secondary GTCS as the only seizure. Eye opening, hypermotor seizure, bilateral asymmetric clonic seizure, vocalization, and contralateral body turning occurred more frequently in the MRI- group compared to the MRI+ one. MRI- patients share some semiological features with FCD as opposed to HS patients. Conclusion. MRIepilepsy patients frequently present with electroclinical features seen in frontal lobe epilepsy or in epilepsy associated with FCD.


Introduction
Epileptogenic lesions carry intrinsic epileptogenicity or epileptogenic potential in their close vicinity.Pharmacoresistant focal epilepsy with well-defined cortical lesions, owing to improved acquisition and interpretation of brain magnetic resonance imaging (MRI+) is a subject of surgical treatment frequently leading to favorable outcome.However, there exists a significant proportion of patients with MRI not showing a lesion potentially causative of chronic epilepsy.Patients studied presurgically with MRI negative (-) focal epilepsy and epileptogenic zone potentially located anywhere in the cerebral cortex account for 18% to 43% [1][2][3] .In theory, in those patients an epileptogenic zone could be located anywhere in the cerebral cortex.Nevertheless, some authors found a high prevalence of frontal lobe epilepsy in a group of consecutively recruited MRI(-) refractory epilepsy patients 3 .
The analysis semiology of symptoms during epileptic seizures helps to determine the epileptogenic zone.Some studies of seizure semiology helped to differentiate between seizures arising in the frontal region from the mesial temporal regions [4][5] .Although MRI(-) patients represent a significant subgroup of epileptic patients its overall semiology is not extensively studied.In the setting of long-term video-EEG monitoring (vEEG) we investigated seizure semiology and EEG findings in MRI(-) and MRI(+) epileptic patients with the premise that they may differ.

Methods
Patients were selected from the database of the Epilepsy Center vEEG Monitoring Unit at the Neurology Clinical Center of Serbia covering the period from August, 2009 till May, 2012.We used the Vyasis Nicolet 64-channel acquisition system with 10-20 electrode placement system with anterior temporal electrodes added.Antiepileptic drugs (AEDs) were discontinued in the absence of patient-specific contraindications in all patients, in a well-structured way: 50% of the prescribed daily dose was withdrawn upon admission; for patients on polytherapy the complete withdrawal of one drug was favored.We studied 180/310 patients (58.04%) who underwent long-term vEEG monitoring in whom focal epileptic seizures were recorded.One hundred thirty patients were excluded due to focal epilepsy with isolated auras (65 patients), psychogenic nonepileptic seizures (30 patients), generalized epilepsy (28 patients), brain MRI not performed (3 patients), syncope (3 patients) and epilepsia partialis continua (1 patient).
In patients with multiple seizures recorded, one was taken as a "reference" (defined as the best seizure seen on video, with secondary generalized tonic-clonic (sGTC) phase if this occurred).Semiological signs recorded only in other (non-reference) seizures were added according to time of the occurrence in the final sequence of semiological signs of the "reference" seizure.Clinical onset of the seizure was determined by the first visible change in behavior, or when the patient announced an aura or pressed the seizure alarm.Data concerning quality of auras were obtained during the aura, immediately at the end of the seizure or from the medical history.We defined the onset of generalization as the brief period between the focal seizure and the remaining sGTC phase characterized by head version or vocalization (6).Determination was used as the cut-off point for analysis.
We analyzed seizure duration in all patients.Seizure duration was analyzed in the "reference" seizure if multiple seizures occurred.Seizure onset was defined as the first ictal EEG change or first subjective/objective clinical change, whichever happened first.Ictal EEG was defined as rhythmic electrical activity with evolution in frequency, amplitude and distribution.Ictal onset zone (ictal EEG), and location and number of the interictal epileptiform abnormality populations were included in the analysis.
The ictal characteristics listed in appendix e-1 5 were looked for in every seizure.Lateralization of the semiological signs was determined according to the side of the lesion or ictal EEG.
Descriptive statistics including numbers and percentages of categorical variables or mean, median, and standard deviation were used to characterize the study sample.Differences between groups in number and frequency of semiological signs were analyzed using Student's t-test, one-way ANOVA followed by Tukey's post hoc test and Pearson chi-square test or Fisher exact probability test when appropriate.Temporal relations between most frequent symptoms were analyzed using log-linear models.Hierarchical cluster analysis with Furthest neighbour and Euclidean distance metric were used to describe clustering of semiological signs which were included in analysis only if frequency was >20%.Statistical analyses were performed using SPSS for Windows, version 22.The significance level was set at 0.05 in all analyses.The research was approved by the Ethical Committee of the Clinical Center of Serbia.Written informed consent was obtained from all patients.
Average duration of long-term vEEG monitoring was 3.98 days (range 2-4 days).A total of 1712 seizures were recorded (median 5 seizures per patient, range 1-384 seizures).A single seizure was recorded in 11.1% of patients.A total of 184 sGTC seizures were recorded (median 2 seizures per patient, range 1-10 seizures) in 74 patients (41.1%).In 12.7% of patients only sGTC seizures were recorded (single sGTC seizure in 5%).Secondary GTC seizures as the only seizure type were more likely in MRI(-) patients compared to patients with HS or FCD (p<0.001) (Figure 1).
Results indicated a significant difference for various two-sequential signs that appeared anywhere in seizures in MRI(+) and MRI(-) groups.Frequencies are reported in Table 1.
As the first symptom epigastric aura was more common in HS group and eye opening in MRI (-) and FCD groups (p<0,001).First three most common sequential symptoms and signs were different in analyzed groups (Table 2).
Cluster analysis yielded patterns of symptom grouping that were different for MRI(-) group, HS group, and FCD group (Figure 2).Two major clusters of signs appeared in the majority of seizures in MRI (-) group (Figure 2A): contralateral tonic-clonic arm seizure or GTC seizure associated with eye opening, and oroalimentary automatisms with behavioral arrest.Similarly, in FCD group two clusters of signs emerged (Figure 2B): isolated and less frequent GTC seizure cluster, and complex cluster consisting of eye opening and cephalic aura closely associated with smiling/laughing, unilateral hand automatisms and head turning, or more distant oroalimentary automatisms with behavioral arrest.Four well demarcated cluster signs appeared in the HS group (Figure 2C): epigastric aura with behavioral arrest and oroalimentary automatisms as a frequent cluster was associated with the cluster of similar frequency consisting of contralateral arm immobility or hand dystonia, unilateral head or body turning and arm automatisms; the cluster consisting of staring, vocalization and contralateral clonic arm seizure was associated with a less frequent cluster of GTC seizures.

Discussion
For seizure expression the site of etiological lesion is the most critical even if its electroclinical presentation appears to be remote 7 .However, semiology and EEG findings remain major objective measures that help delineate the epileptogenic zone in pharmacoresistant focal epilepsy patients with normal MRI findings.In this study we contrasted overall electroclinical features of MRI(-) patients to the MRI(+) group and its two distinctive representationspatients with HS and FCD.Although several studies analyzed different aspects of MRI(-) patients 1, 3, 8-9 , we focused on electro-clinical characteristics of patients with focal epilepsy and negative MRI findings.
Our MRI(-) patients to a certain extent have similar features seen in frontal lobe epilepsy (FLE).We found a common presence of the sGTC seizures as an exclusive seizure type in MRI (-) patients.A very high percentage of sGTC seizures (50-90%) was reported in studies of FLE (10-13).Furthermore, single photon emission computed tomography (SPECT) images of the GTC seizure during electroconvulsive therapy revealed the greatest signal increase in the frontal and parietal cortices 14 .Furthermore, MRI(-) patients exhibit a significant absence of the interictal epileptiform abnormalities.This could be due to the generation of spikes in deep extratemporal cerebral tissue (midline or supraorbital frontal cortices) as discussed in a previous report 15 .Finally, in our group MRI(-) compared to MRI(+) patients had different ictal EEG onset localization ie.MRI(-) had more common ictal onset zone in the frontal or frontocentral region.
Some clinical features commonly occurred in MRI(-) patients.Eye opening as an early sign occurs in nocturnal seizures typically associated with FLE (16).Similarly, hypermotor seizures, and ictal vocalization are a common consequence of the symptomatogenic zone activity in different frontal lobe regions [17][18] .Still, the significance of contralateral body turning and bilateral asymmetric clonic seizure in MRI(-) patients is not clear.Bilateral asymmetric clonic seizure corresponds to the pretonic phase of the GTC seizure.GTC seizures whose clinical heterogeneity suggest that full expression is less common than fragmentary events (6).Therefore, bilateral asymmetric clonic seizure may represent a distinctive phase of the GTC seizure in MRI(-) patients with uncertain localizing value.
In our cohort the proportion of MRI(-) patients and distribution of detected cortical etiology in the MRI(+) group (highest prevalence of HS) are comparable with earlier findings 2, 19.Therefore, signs commonly present in the MRI(+) group correspond to those habitually seen in mesial temporal lobe epilepsy.Similarly, our results of the twosequential signs that appeared anywhere in the seizures are those that appear in temporal lobe epilepsy patients [20][21] .
In order to better understand the clinical characteristics of our MRI(-) patients we compared semiological differences with HS and FCD subgroups, which are the most common pathologies seen in large cohorts (19).Patients in the MRI(-) group significantly differ from HS patients.This finding is consistent with the hypothesis that MRI(-) patients most commonly do not have mesial temporal lobe epilepsy.Interestingly, FCD patients share similar differences from HS patients as MRI(-) patients.In contrast, the MRI(-) and FCD group distinction is not that noticeable.Cluster analysis indicates two major clusters of signs in MRI(-) and FCD groups, and three major clusters of signs in the HS group.One cluster is identical in all analyzed groups, with signs depicting sGTC seizure.It is worth noting that the secondary GTC seizure subcluster is the most complex in the MRI(-) group.The HS group showed a distinct cluster of signs (epigastric aura, oroalimentary automatisms and behavioural arrest, or contralateral hand immobility/dystonia, unilateral arm automatisms, unilateral head and body turning), well described as typical in mesial temporal lobe epilepsy 4 .Although MRI(-) and FCD group share some common signs in the remaining subcluster, these are not equally distributed.This finding can be attributed to the limitation of the cluster analysis dendogram presentation; rearranging the ordering of symptoms in a dendogram can to some degree be feasible without having an effect on the meaning of the diagram 5 .Our results indicate that MRI(-) patients present some electro-clinical features seen commonly in FLE.Further, some of the clinical characteristics of MRI(-) patients resemble those reported in epilepsy associated with FCD.It seems possible that higher-resolution MRI (i.e. higher field-strength magnet with thinner slices), may increase the chances to detect small frontal lobe lesions not seen in our scanning protocol.However, a well-defined clinical syndrome in the MRI(-) group remains to be identified.Our study improves knowledge about focal MRI(-) epilepsy.Further research comparing MRI(-) epilepsy and larger sample of FCD patients, should establish the presence of a clinically distinct entity in focal epilepsy patients with no epileptogenic lesion.

a
Patients with HS were significantly older than MRI(-)patients (p=0.037);b There was no difference across variables; c Epilepsy duration in HS patients was significantly longer than in MRI(-)patients (p=0.004) and FCD patients (p=0.008).