Address correspondence and reprint requests to Dr. I. M. Najm at Section of Epilepsy, 9500 Euclid Avenue, S51, Cleveland, OH 44195, U.S.A. E-mail: firstname.lastname@example.org
Summary: Purpose: Malformations due to abnormal cortical development (MCDs) are common pathologic substrates of medically intractable epilepsy. The in situ epileptogenicity of these lesions as well as its relation to histopathologic changes remains unknown. The purpose of this study was to correlate the cellular patterns of MCDs with the expression of focal cortical epileptogenicity as assessed by direct extraoperative electrocorticographic (ECoG) recordings by using subdural grids.
Methods: Fifteen patients with drug-resistant focal epilepsy due to pathologically confirmed MCD who underwent subdural electrode placement for extraoperative seizure localization and cortical mapping between 1997 and 2000 were included in the study. Areas of interictal spiking and ictal-onset patterns were identified and separated during surgery for further pathologic characterization (cellular and architectural). Three pathologic groups were identified: type I; architectural disorganization with/without giant neurons, type IIA; architectural disorganization with dysmorphic neurons, and type IIB; architectural disorganization, dysmorphic neurons, and balloon cells (BCs). The focal histopathologic subtypes of MCDs in cortical tissue resected were then retrospectively correlated with in situ extraoperative ECoG patterns.
Results: Cortical areas with histopathologic subtype IIA showed significantly higher numbers of slow repetitive spike pattern in comparison with histopathologic type I (p = 0.007) and normal pathology (p = 0.002). The ictal onset came mainly from cortical areas with histopathologic type IIA (nine of 15 patients). None of the seizures originated from neocortical areas that showed BC-containing MCD (type IIB).
Conclusions: This study shows that areas containing BCs are less epileptogenic than are closely located dysplastic regions. These results suggest a possible protective effect of BCs or a severe disruption in the neuronal networks in BCs containing dysplastic lesions. Further studies are needed to elucidate the nature and the potential role(s) of balloon cells in MCD-induced epileptogenicity.
Malformations due to abnormal cortical development (MCDs) are frequent causes of medically intractable epilepsy that is amenable to surgical resection. But the presence of MCD is not always associated with a history of epilepsy (1,2). Pathologic studies on human brain tissues surgically resected from patients with a history of drug-resistant focal epilepsy identified various architectural and cellular changes (1–4). The main histologic hallmark of MCD is the presence of columnar and laminar disorganization that can be intermixed with various cellular abnormalities that include dysmorphic neurons, giant neurons, and balloon cells (BCs). Recent studies on patients who underwent surgical resection for the treatment of drug-resistant epilepsy suggested characteristic clinical, imaging, and electrocorticographic (ECoG) patterns that are correlated with the presence or absence of BCs (large cells with eccentric nuclei and opalescent cytoplasm). Patients with BC-containing MCDs had an earlier age at seizure onset, more frequent seizures, and worse postoperative outcome (5–9). Moreover, BC-containing MCDs are characterized by signal increase on fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI) sequence (10). With histopathologic data acquired from patients who had direct ECoG recordings and underwent en bloc surgical resection, Rosenow et al. (1998) showed a significant increase in the number of spikes in BC-containing MCD as compared with MCD that was devoid of BCs (9). However, no in situ and direct correlations were made between the interictal EcoG spiking and the various histopathologic patterns that are usually found in the same patient (11,12).
In this study, we directly correlated the different histopathologic subtypes of MCDs with the frequency and patterns of interictal spiking and ictal EcoG patterns acquired in patients who underwent prolonged extraoperative ECoG recordings by using subdural electrode recordings.
MATERIALS AND METHODS
Fifteen patients with medically intractable focal epilepsy secondary to MCDs, who underwent presurgical evaluation with prolonged surface EEG and extraoperative invasive recordings with subdural grids between 1997 and 2000, were included in the study. Medical records were reviewed to identify potential risk factors for the development of epilepsy. These potential risk factors include prenatal trauma, perinatal bleed or infection, complicated delivery, childhood febrile seizures, head trauma, or family history of seizures. Preoperative MRI studies that included volume acquisition T1-weighted and FLAIR sequences were performed in all patients included in the study.
All patients underwent surgical resection of the MCD lesion with direct pathologic correlations. The recommendation for invasive subdural grid electrodes monitoring was taken during the weekly Cleveland Clinic Epilepsy Center patient-management conference. As recently described, the two main driving issues behind the decision for invasive monitoring are (a) definition of the epileptogenic zone, and/or (b) the localization of the functional (eloquent) cortical regions and its (their) relations with the epileptogenic area(s) (13). In selected cases, intraoperative ECoG was done only in the setting of intraoperative functional electrical stimulation.1 Craniotomies were done, and large subdural grids were placed. Contiguous covering of the neocortical areas was achieved through the placement of adjacent grids, with the goal of not leaving any intervening unsampled cortex. No strips were placed. All subdural electrode placements and surgical resections were performed by the same neurosurgeon (W.B.). The demographics and clinical/imaging/pathologic/seizure outcome data on the patients included in the study are shown in Table 1. Patients with associated tumor or vascular malformation on pathologic examination were eliminated from the study. The study was approved by the Institutional Review Board of the Cleveland Clinic Foundation.
Localization of subdural electrode positioning by using MRI-based three-dimensional reconstruction
All 15 patients had MRI studies within 24–48 h of subdural electrode implantation. Three-dimensional surface reconstruction of patient's brain was created to provide visual correlation between each subdural electrode position and corresponding cortical area. As previously described, T1-weighted MRIs were acquired and combined to form a single three-dimensional volume with volume element dimensions of 0.9 × 0.9 × 2.0 mm (11). Volumes were processed to remove background elements and to segment the brain automatically. Electrodes on the two-dimensional MRIs were semiautomatically segmented and saved with the segmented brain volume. The surface of the brain was reconstructed by using an in-house computer program that interactively renders the volume of the brain and the surface location of the electrodes. A red 2-mm diameter sphere was surface rendered at the center of each segmented electrode and overlaid on the volume-rendered images. This reconstruction allows accurate identification of electrode position for surgical resection of epileptogenic tissue.
Tissue characterization and histopathologic studies
As previously reported (11), in all patients, the exact locations of the subdural electrodes that showed the presence of interictal spiking, and those showing ictal onset patterns (as defined later) were intraoperatively identified by use of sterile brilliant green by the neurosurgeon (W.B.). Three patients underwent hippocampal and temporal neocortical resections after prolonged video-invasive EEG recordings. All three patients showed evidence of “dual pathology” that consisted of hippocampal sclerosis (HS) and MCD. The resected blocs varied in size (depending on the number of involved subdural electrodes). The resected tissue was taken from a midline point between the two electrodes recording distinct electrical activities (∼0.5 cm from the center of the adjacent electrode). Representative parts of the specimens were submitted to the Department of Pathology of the Cleveland Clinic Foundation for independent pathologic interpretation, and other adjacent cubes were processed for direct histopathologic correlations.
Cubes of cortical specimens were labeled and immersion-fixed for ≤36–48 h in 4% paraformaldehyde at 4°C, and then cryoprotected with 20% buffered sucrose for processing as described later. Because the resected tissue was not grossly abnormal in most cases (e.g., with macrogyria or polymicrogyria), and the dysplastic regions were not always detectable by visual observation, the representative cortical cubes were coronally oriented for histologic staining. At least three sections (30 μm thick) at alternating 1-mm intervals from each cube were collected for staining with cresylecht violet (CV). Various neocortical regions were analyzed at ×5 and ×40. The presence of focal MCD was confirmed independently by three of the authors (R.P., I.N., and Z.Y.) on CV staining in at least one of the resected samples from each patient included in the study. The following histopathologic features were characterized: (a) cortical architecture: laminar organization, columnar arrangement, the persistence of neurons in the molecular layer (layer I), and the presence of neurons in the subcortical white matter; and (b) cellular morphology: neuronal orientation, CV staining intensity, and for the presence of large cells with central nuclei and well-defined cytoplasmic membranes (giant neurons or meganeurons) and BCs (strikingly large opalescent cytoplasm with eccentric nuclei). The diagnosis of MCDs was made on identification of disorganization in the cortical architecture (dyslamination and columnar disorganization) and the presence of dysmorphic neurons, giant neurons with central nuclei, and/or BCs. As shown in Fig. 1, based on the histopathologic subtypes, the resected tissue was divided into three groups (14):
Type I: Presence of architectural disorganization of the cortical layers with or without giant neurons (Fig. 1B,F)
Type IIA: Presence of architectural disorganization intermixed with dysmorphic neurons (Fig. 1C,G)
Type IIB: Presence of architectural disorganization and BCs (Fig. 1D,H)
Electrocorticographic recordings and patterns
The ECoG data were acquired and saved by using the Vangard digital EEG recording system according to the following settings: (a) sampling rate of 200 Hz, (b) low filter setting of 1 Hz and high filter setting of 70 Hz, (c) the analyzed samples consisted of 2-min epochs that were saved per hour of continuous monitoring. With referential recordings, the EcoG data acquired from the prolonged extraoperative subdural electrodes corresponding to each of the resected cortical areas were retrospectively and independently analyzed by three of the authors (K.B., G.K., and I.N.). In the case of disagreement between the three authors reading the ECoG data, the disagreement was resolved after group review, and a final decision is made. As illustrated in Fig. 2, the interictal EcoG patterns were divided into four groups:
1Isolated spikes: amplitude >200 μV, frequency >7 Hz, irregular firing,
3Paaroxysmal fast: burst duration >0.5 s, frequency >10 Hz, regular firing, and
4Runs of slow repetitive spikes: burst duration >0.5 s, amplitude >200 μV, frequency <7 Hz, regular firing
The first 20 min of interictal ECoG recordings that were digitally saved for each patient were analyzed. The 20 min consisted of different sample files of 2 min long, which were selectively saved during the evaluation (by a monitoring epileptologist) as containing epileptic activity because they were representative of various ECoG patterns. As it was difficult to characterize the state of arousal/sleep of the patient in the saved recordings (there were no concomitant scalp recordings in most of the patients), only the sample files saved between 6.00 a.m. and midnight were used, to minimize variation of the spikes frequency between various times of the day.
The frequencies of the ECoG interictal spiking (per minute and per hour) and their patterns were analyzed for each subdural electrode. As for the repetitive spiking, and paroxysmal fast pattern, each burst of repetitive spiking and paroxysmal fast pattern was counted as one unit of interictal spiking activity. Thereafter the data acquired from all electrodes that sampled patterns recorded from cortical area(s) showing the same histopathologic characteristics as defined earlier were grouped and averaged. For the purpose of data analyses, each patient had only one ECoG data point for each interictal pattern and for each histopathologic subgroup. As illustrated in Fig. 3, the ictal-onset patterns were divided into three groups:
1Paroxysmal fast: duration >10 s, amplitude >50 μV, frequency >10 Hz, with evolution in amplitude and/or frequency,
2Repetitive spiking: duration >10 s, amplitude >200 μV, frequency 3–10 Hz, with evolution in amplitude and/or frequency, and
3Paroxysmal fast with repetitive spiking: duration >10 s, amplitude >50 μV, frequency >10 Hz, intermixed with repetitive spiking and with evolution in amplitude and/or frequency.
The median value of the frequency of each different type of interictal spikes per hour (isolated spikes, repetitive spikes, paroxysmal fast activity, and runs of slow repetitive spike patterns) seen at electrodes overlying cortical area with the histopathologic subtype I, IIA, IIB, and normal in each patient were compared by using the nonparametric Kruskal–Wallis test for overall association and Dunn's procedure for pairwise comparison among groups, adjusting the significance criterion with a Bonferroni correction within each outcome. Because the data was not normally distributed and both the number of the subjects and number of observations for each pathology type were small, an analysis of variance adjusting for the likely within-subject correlation was not appropriate.
The frequency of each interictal spike pattern (isolated spikes, repetitive spikes, paroxysmal fast activity, and runs of slow repetitive spikes) per each electrode were compared between each different pathologic subtypes (I, IIA, IIB, and normal) by using a version of nonparametric Wilcoxon test that would adjust for the correlation within the subject. The technique called bootstrap also was used to obtain a more accurate estimation of the standard error of the mean rank difference for each pair of pathologies. A significance criterion of p < 0.0083 was used for multiple comparison.
As shown in Table 1, at time of surgery, the patients' ages ranged between 6 months and 34 years. The age at seizure onset was between 3 weeks and 20 years. Seizure frequencies varied from two to three per year to a maximum of 100 per day. Seizure semiology varied depending on the location of epileptogenic area. Twelve patients were seizure free at last follow-up (>1 year). Two of the three patients who continued to exhibit seizures had partial resection of the ictal-onset zone as it overlapped with the primary motor areas (patients 2 and 9). Patient 14 had recurrence of the seizures but with a new semiology (new seizures are automotor) with the onset region localized after remonitoring (scalp EEG) to the temporal lobe.
MRI and scalp EEG findings
Focal MRI abnormalities that corresponded to the areas of confirmed histopathologic changes were seen in 10 (67%) patients: four of 10 patients had focal MRI abnormalities in the frontal area, two of 10 patients showed abnormalities in the precentral gyrus, and three of 10 patients had MRI lesions localized to the temporal lobes. Patient 3 had focal MRI abnormalities in the frontal and temporal regions. In three of 10 patients (patients 3, 8, and 15), MRI showed hippocampal atrophy in addition to other adjacent focal neocortical abnormalities.
Interictal scalp EEG recordings showed regional or multiregional (more than three nonadjacent lobes) sharp waves in six (40%) of 15 patients, and three (20%) patients had generalized sharp waves.
Detailed pathologic examinations of various resected cortical cubes from all 15 patients included in the study showed the following: Two patients had isolated type I MCD, and four patients had isolated IIA changes. Two patients had distinct areas with types I and IIB changes, four patients had distinct cortical regions with types I and IIA MCDs, and two patients had distinct IIA and IIB regions. One patient had pathologic changes of types I, IIA, and IIB in three separate (but adjacent) resected cortical areas (patient 2). Dual pathology (HS and MCD) was identified in three patients (3, 8, and 15).
Electrocorticographic and pathologic correlations
As shown in Table 2, electrodes overlying type IIA MCD regions showed higher total spikes and isolated spikes than did electrodes sampling type IIB areas, although this did not reach statistical significance. Slow repetitive spike frequency was significantly higher in type IIA than in “normal” regions (p = 0.002) and areas with type I changes (p = 0.007, significant criteria p < 0.0083).
Table 2. Mean and median spike frequencies for each spike pattern as correlated with the corresponding histopathologic change
Normal (n = 5)
I (n = 9)
IIA (n = 11)
IIB (n = 5)
Median (p25, p75)
Median (p25, p75)
Median (p25, p75)
Median (p25, p75)
Statistically significant comparing type IIA and normal (p = 0.002), type IIA and type I (p = 0.007; significant criteria p < 0.0083).
n, number of patients with this type of pathology; SD, standard deviation; p25, 25th percentile; p75, 75th percentile; RS, repetitive spike; PF, paroxysmal fast; SRS, slow repetitive spike pattern; h, hour.
A total of 163 EEG seizures was recorded and analyzed. The average number of seizures that were recorded for each patient was 11.73 (median, 11 seizures). The total number of seizures ranged from one (patient 1) to 30 seizures (patient 5). The total number of seizures recorded during the period of monitoring did not appear to have any correlation with a particular histopathologic subtype. As shown in Table 3, the ictal onset was localized to IIA regions in the majority of patients included in the study (nine of 15; 60%). Ictal-onset zones were recorded from electrodes overlying type I MCD in three (20%) patients. The electrodes underlying the parahippocampal gyrus showed ictal-onset patterns in the three patients who had pathologically confirmed HS. None of the seizures originated from neocortical areas that showed BC-containing MCD (type IIB). One patient (patient 1) had only one atypical aura recorded during the period of invasive EEG monitoring with no EEG changes (the patient had frequent interictal spikes that helped to map the potentially epileptogenic area and its anatomic relation with the eloquent cortical areas). The most commonly seen ictal-onset ECoG pattern in cortical area with histopathologic subtype IIA was paroxysmal fast with or without repetitive spikes (six of nine patients with type IIA pathology and both patients with HS). The ictal-onset pattern seen in MCD type I showed only repetitive spikes with no paroxysmal fast pattern (three of three patients). The ictal-onset zone was colocalized with the interictal areas showing slow repetitive spikes (SRS) pattern in 14 of the total of 15 patients.
Table 3. Correlations between ictal electrocorticographic patterns and histopathologic characteristics
Normal (n = 5)
I (n = 9)
IIA (n = 11)
IIB (n = 5)
PF, paroxysmal fast; RS, repetitive spike.
(n = 0/5)
(n = 3/9)
(n = 9/11)
(n = 0/5)
PF ± RS
Our study is the first to demonstrate a direct in situ correlation between various types of MCDs and both interictal and ictal epileptic patterns in patients who underwent cortical resection for the treatment of medically intractable focal epilepsy. Our results show a differential expression of in situ epileptogenicity in selected types of MCDs. Ictal-onset patterns were not found in BC-containing regions (type IIB) despite the presence of severe cortical architectural disorganization and a high number of dysmorphic neurons. The most severe interictal and ictal in situ changes were seen in MCDs that are characterized by the presence of dysmorphic cells and architectural disorganization (type IIA).
Our results are in agreement with those reported by Munari et al. (15) by using depth EEG recordings. In one of the two cases reported by these authors, the ictal-onset area was mainly recorded from the cortical area surrounding the lesion that was shown to contain BCs on histopathologic examination (15). These results are in apparent disagreement with a recent report on the epileptogenicity of cortical dysplasias (16). With retrospective analyses of resected tissue from patients with medically intractable focal epilepsy who underwent stereotactic depth EEG monitoring (SEEG) over a long period that preceded CT or MRI scanning, Chassoux et al. (16) reported recording of epileptogenicity from BC-containing lesions. The majority of the patients included in that study had BCs (24 patients of a total of 28 patients: 85.7%) as compared with other series reporting on a smaller proportion of MCDs with BCs (5,9). The methods of the two studies are significantly different: (a) We performed a prospective study that included patients with high-resolution MRI and direct electrode-anatomic correlations, whereas the recent report is based on a retrospective review of resected samples with no preoperative MRI correlates on most of the patients; (b) we assayed epileptogenicity by using surface cortical recordings (subdural grids), whereas the epileptogenicity in the report by Chassoux et al. was assessed by depth electrodes.
Our group previously reported on an increased epileptogenicity in patients with BC-containing dysplastic lesions (9). In the previous study, direct correlations between the in situ histopathologic changes and ECoG findings were not made through an intraoperative identification of various areas as defined by imaging and/or ECoG findings. In the current study, we report on the direct correlations between various histopathologic subtypes (that may be present in the same patient) and in situ ECoG patterns. Therefore interictal activities that may be generated in non–BC-containing but dysplastic lesions may have been attributed to BC-containing MCDs.
Our findings support the previous concept of the presence of intrinsic epileptogenicity in some forms of MCDs. The high degree of in situ epileptogenicity of focal MCDs has been observed in previous studies from a large number of patients with intractable epilepsy who underwent invasive/direct cortical EEG recordings (7,8). Moreover, patients with focal MCDs have high seizure frequency and increased incidence of status epilepticus (6–9,17). By using direct intraoperative ECoG recordings, Palmini et al. (8) demonstrated the selective occurrence of distinct in situ occurrence of ictal or continuous epileptogenic discharges (I/CEDs) in focal cortical dysplasias as compared with control epileptic pathologies (such as tumors or arteriovenous malformations). Our results extend these findings, as we show that the expression of in situ epileptogenicity in focal MCDs is differentially correlated with specific types of histopathologic neocortical changes.
With the same methods, we recently showed that BC-containing dysplastic lesions are nonfunctional on direct cortical electrical stimulation, and they show significant FLAIR increased signal on MRI (12). The lack of cortical eloquent function in BC-containing regions could be secondary to severe architectural disorganization and subsequent disruption of the neuronal circuits in those cortical areas. The same mechanism(s) may underlie the decreased in situ epileptogenicity, as the build-up of neuronal synchrony is impaired despite the presence of hyperexcitable dysmorphic neurons. Cellular and architectural studies are needed to elucidate the possible mechanism(s) of lack of function and epileptogenicity in the BC-containing MCDs.
The role and function of BCs in the setting of MCDs is not yet known. Previous studies showed that these large opalescent cells with multinuclear and eccentric nuclei have neuronal and glial characteristics (5). We recently acquired preliminary evidence on the presence of an increased protein density of glutamate-clearance mechanisms (glutamate transporters and glutamine synthetase) within the BCs and in BC-containing regions (Najm et al., unpublished data). These glutamate-clearance mechanisms may be limiting the spread of epileptogenicity in these lesions. Studies to confirm these findings and to assess their significance are under way at our laboratory.
Our study shows that some interictal epileptic patterns may predict the pathologic characteristics and the location of ictal-onset zones. According to our findings, certain patterns of interictal epileptiform activities correlated with specific histopathologic subtypes (as proposed by Palmini and Lüders) (14). Cortical areas that were characterized by architectural disorganization and the presence of dysmorphic neurons (subtype IIA) showed higher prevalence of slow repetitive spiking as compared with type I regions. The areas that showed slow repetitive spiking often overlapped with ictal-onset zones in areas containing dysmorphic neurons (subtype IIA).
Further studies are needed to validate these results through the analyses of postsurgical seizure outcome and to establish the basic pathophysiologic mechanism(s) responsible for decreased epileptogenicity in the BC-containing regions.
Acknowledgment: This study was supported by NIH Grants K08 NS02046 and R21 NS42354 to IN.