Statistical analysis was conducted by Hélène Mellerio, MD, Unite d’Epidémiologie Clinique – Inserm CIE-5, Hôpital Robert Debré, Paris, France.
Address correspondence to Francine Chassoux, MD, Service de Neurochirurgie, Centre Hospitalier Sainte-Anne, 1, rue Cabanis, 75014 Paris, France. E-mail: firstname.lastname@example.org
Purpose: Type II focal cortical dysplasia (TTFCD), a highly epileptogenic lesion with severe epilepsy curable by surgery, is missed by magnetic resonance imaging (MRI) in about one third of cases. Little is known about the electroclinical presentation in these MRI-negative patients and a poor surgical outcome is frequently reported. We compared the clinical and neurophysiologic features in MRI-negative and MRI-positive cases in order to better identify candidates for surgery.
Methods: Among 62 consecutive TTFCD patients (38 male, 24 female; 7–52 years old; 22 children) operated for intractable epilepsy, 25 (40%) presented negative MRI findings. We compared the history of epilepsy; the type, frequency, and distribution of seizures; neurologic examination cognitive and psychiatric impairment; interictal-ictal electroencephalography (EEG) and stereo-EEG (SEEG) data, fluorodeoxyglucose positron emission tomography (FDG-PET) data, neuropathologic findings; and surgical outcome in the MRI-negative and the MRI-positive groups.
Key Findings: Severe partial epilepsy beginning in childhood, high seizure frequency including status epilepticus, stereotyped seizures suggestive of precise brain localization, extratemporal location and functional area involvement were characteristic and similarly found in both groups. On EEG, pseudorhythmic activity was found in about 40% of patients in each group. SEEG recordings demonstrated the typical pattern characterizing TTFCD in both groups. FDG-PET had a localization value in 84% of the MRI-negative cases and helped to delineate the dysplastic cortex in 65% of the MRI-positive cases. The combination of imaging and neurophysiologic data allowed us to perform safe and restricted resections, limited to a single gyrus in more than half of all cases. In addition, we were able to avoid invasive monitoring in most MRI-positive cases and even in some selected MRI-negative cases. The proportion of patients with a favorable surgical outcome was comparable in both groups (88% in MRI-negative and 94% in MRI-positive cases). The main difference between the groups was a significantly higher frequency of sleep-related epilepsy in the MRI-negative group (p = 0.028). This phenotypic characteristic provides a new argument for TTFCD in MRI-negative extratemporal epilepsy.
Significance: These results lead us to consider that children or adult patients in whom electroclinical data suggest TTFCD, are highly suitable for surgery, especially for cryptogenic sleep-related epilepsy.
Despite the efforts made during the last 10 years to classify focal cortical dysplasias, among which TTFCD corresponds to severe forms, including FCD with dysmorphic neurons (type IIA) and balloon cells (type IIB) (Palmini et al., 2004; Blümcke et al., 2011), few data are available on the clinical and neurophysiologic characteristics associated with MRI-negative TTFCD. The aim of the present study is to describe the electroclinical features in this cryptogenic epilepsy population in order to better identify potential candidates for surgery. To this end, we compared the MRI-negative patients to MRI-positive patients (i.e., patients with typical MRI features), in terms of their demographic and electroclinical characteristics and surgical outcome, in a recent series of patients operated for intractable epilepsy due to TTFCD.
Patients and Methods
Sixty-two (17%) of 358 patients with intractable epilepsy who underwent a corticectomy in our institution during the last 10 years had histologically proven TTFCD and constituted the study population. The population consisted of 38 male and 24 female patients, including 22 children. Age at the time of surgery ranged from 7 to 52 years (mean 22.8 years), age at epilepsy onset ranged from the first days of life to 20 years (mean 7 years), and the preoperative duration of epilepsy ranged from 1–45 years (mean 15.8 years). A history of severe partial epilepsy with high seizure frequency was reported in all cases and 54 patients (87%) presented with daily seizures (up to 30 a day) when referred for surgery. Antenatal or perinatal adverse events were reported in five patients and febrile seizures in two.
Presurgical evaluation consisted of ictal video–electroencephalography (EEG) (including sleep recordings in all patients with predominantly nocturnal seizures), high-resolution MRI, functional MRI (fMRI), and fluorodeoxyglucose positron emission tomography (18FDG–PET) scans for all patients. Thirty-seven patients (60%) also underwent stereo-EEG (SEEG), as described previously (Bancaud, 1980; Talairach et al., 1992; Chassoux et al., 2000, 2007). We collected data from the patients records on the following: history of epilepsy; type, frequency, and distribution of seizures (predominantly nocturnal and sleep-related seizures were reported by the patients and their families and assessed by observation during the EEG monitoring); neurologic, cognitive, and psychiatric examinations; interictal-ictal EEG and SEEG data; MRI and PET imaging features; neuropathologic findings; and surgical outcome. Brain MRIs were performed on a 1.5 Tesla Magnet (Signa Excite; General Electric Healthcare, Milwaukee, WI, U.S.A.), using as receiver coil an 8HRBRAIN coil and including the following sequences: three-dimensional (3D) gradient echo T1-weighted inversion recovery acquisition [slice thickness = 1.2 mm, field of view (FOV) = 240 × 240 mm, 256 × 192 matrix, 1 Number of excitations (Nex)], two dimensional (2D) coronal fast spin echo T2-weighted acquisition (thickness 4 mm, no gap, FOV = 240 × 240 mm, 512 × 256 matrix, 2 Nex) and fluid attenuated inversion recovery (FLAIR) using 2D axial or coronal contiguous slices (thickness 5 mm, FOV = 240 × 240 mm, matrix 256 × 192, 1 Nex). Motor fMRI was systematically used in patients with TTFCD located in the central region. PET scans were performed in five patients using a head-dedicated PET camera with 5.8 mm in-plane and 5 mm axial resolution (ECAT 953/31B Siemens) and in all other patients using a 3D camera allowing axial sampling of 2.46 mm (HR+ CTI Exact Siemens, Knoxville, TN, U.S.A.). PET and MRI images were routinely coregistered using ANATOMIST software (SHFJ, CEA, Orsay, France). PET examination was obtained as part of a research protocol approved by the local ethical committee, and written informed consent was obtained for all patients. All MRI studies were reviewed by a trained radiologist who specialized in epilepsy, who classified them into two groups: “MRI negative” (normal MRI or nonspecific abnormalities) or “MRI positive” (one or more imaging findings suggesting TTFCD). For the histologic examination, cortical samples were fixed in formalin-zinc and paraffin sections (4 μm) and stained with Hemalun-Phloxin and Nissl-luxol (Kluver-Barera). Complementary immunohistochemistry was done using antibodies directed against glial fibrillary acid protein (GFAP, clone 6F2, DakoCytomation, Carpinteria, CA, U.S.A.), microtubule-associated proteins (MAP2, clone HM-2, Sigma, St. Louis, MO, U.S.A.), and neuronal nuclei (NeuN, clone AYO, Millipore Chemicon, Bedford, MA, U.S.A.). Surgical resections were guided by integration of imaging data, including coregistration of PET with MRI and fMRI and SEEG results, where available. For resections in central areas, motor stimulations were performed during surgery. The extent of surgical resection was classified as gyrectomy if it was limited to a single gyrus or as corticectomy in the other cases. Seizure outcome was assessed according to Engel’s classification (Engel et al., 1993).
Quantitative variables were expressed as median (1st–3rd quartiles: Q1–Q3) and qualitative variables as number (percentage). Within-group comparison was performed using the Wilcoxon rank sum test for nonparametric data. For dichotomous results, a chi-square test or Fisher’s exact test was performed, as appropriate. All statistical tests were two-sided. For multiple testing, a correction was applied according to the Bonferroni rule. A p-value of <0.05 was considered statistically significant. Analyses were computed on SAS (version 9.1) software (SAS Institute, Cary, NC, U.S.A.).
MRI was positive in 37 patients, showing typical MRI features of TTFCD, whereas MRI was classified as negative in the 25 others (40%). Among the MRI-negative cases, subtle abnormalities consisting of unusual sulcus depth or gyral patterns were described in 12 patients, but these findings were regarded as nonspecific or doubtful. In the other 13 patients, the MRI appeared strictly normal. In contrast to the negativity of MRI, a focal or regional hypometabolism was visually detected in 21 cases and in two additional cases after coregistration of PET with MRI (Fig. 1). However, among these 23 positive PET scans, a misleading, predominantly temporal localization was found in two cases of TTFCD, which was located in the orbitofrontal area. Therefore, a correct localization was provided by PET in 21 (84%) of the 25 MRI-negative cases. In the MRI-positive group, PET was positive in 28 cases and considered contributory for TTFCD delineation in 24 (65%), but a false or misleading localization was observed in four other cases (predominantly temporal hypometabolism in orbitofrontal TTFCD in two patients; temporoparietal hypometabolism in a small TTFCD located in the supplemental motor area in one patient; and central hypometabolism in a TTFCD located within the insula in one patient). In the remaining nine cases, no metabolic change was clearly identified, except in the deep part of the sulcus containing the TTFCD; of note, all these patients had a very high seizure frequency and presented several seizures just before or during the PET scan. All but two of these nine patients had a TTFCD located in the central area.
Age at surgery and duration of epilepsy were significantly lower in the MRI-negative group, which had a higher proportion of children, a finding that is explained by a special collaboration with pediatric epilepsy centers, which referred cryptogenic cases to our institution (Table 1).
Table 1. Comparison of clinical, EEG, SEEG, PET, and surgical data in MRI-negative and MRI-positive TTFCD
Clinical data/MRI n = 62
MRI-negative n = 25
MRI-positive n = 37
HD, hemispheric dominance; EPC, epilepsia partialis continua; EZ, epileptogenic zone. Bold indicates statistically significant differences between the two populations.
aIctal recordings available in 31 cases in the MRI-positive group.
bFor MRI-negative cases, lesional extent was based on histological findings.
7–38 years (18)
7–52 years (26)
Children <16 years (n)
Age at onset (min–max/mean)
1–20 years (6.8)
0–20 years (7)
Duration of epilepsy
1–31 years (11)
1–45 years (19)
Characteristics of epilepsy
Antenatal/perinatal adverse events
High seizure frequency (>1/day)
Seizure free interval >1 year
Neurologic deficit (motor)
Low IQ (<80)
FCD location (left/right)
Frontal prefrontal/ premotor
Rhythmic spike activity
PET localization value
EZ = PET/MRI
PET > EZ > MRI
Type of resection
Engel class I – IA (%)
1–10 years (4.6)
1–8 years (3.9)
Balloon cells +/ Balloon cells −
All patients had severe partial epilepsy beginning in childhood or adolescence. Age at onset ranged from the neonatal period (two cases) to 20 years (one case) with a similar mean age of onset (about 7 years) in both groups. Infantile spasms were not reported in any patients. Seizures were stereotyped in all cases, and ictal semiology appeared to be clearly related to the location of the dysplastic lesion. Secondary generalization was either absent or an exceptional occurrence. Severity of epilepsy with a high seizure frequency and occurrence of status epilepticus was similar in both groups. Prolonged seizure-free periods were also seen in both groups (up to 20 years in two MRI-positive cases). Of interest, sleep-related epilepsy (SRE), defined as more than 70% of seizures occurring during sleep (Nobili et al., 2009), was found in 72% of cases in the MRI-negative group and was significantly more frequent than in the MRI-positive group (35%, p = 0.028). This main clinical difference between the two groups was not related to the location of the TTFCD in the frontal lobe, because a similar proportion of MRI-negative patients (52%) and MRI-positive patients (62%) had a frontal TTFCD.
Neurologic deficits tended to be more often observed in the MRI-negative group (20% had a mild motor deficit). Left-handed patients were more frequent in the MRI-positive group (27% vs. 12%), with atypical language dominance in three of them. However, these clinical differences did not reach the threshold of significance. Cognitive impairment with low IQ (64–76) was found in a minority of patients (16%, including children and adults) in both groups, but mental retardation (four children) was found only in the MRI-positive group. Psychiatric disorders were not rare and their prevalence was similar in both groups.
Rhythmic spike activity occurring or increasing during sleep was observed on surface EEG in over 40% of cases in both the MRI-negative group and the MRI-positive group (Figs. 2–4). In both groups, interictal and ictal discharges were mainly focal or regional, although diffuse spikes and/or nonlocalized ictal discharges affected a few patients, mostly children (Fig. 5).
SEEG was performed in 21 patients of the MRI-negative group (84%) and in 16 patients of the MRI-positive group (43%). Of note, only one MRI-positive patient underwent SEEG after 2006. Intralesional interictal and ictal activity recorded during SEEG was similar in both groups and the same criteria were used to delineate the dysplastic cortex (Chassoux et al., 2000; Chassoux, 2003) (Fig. 6). In the MRI-negative group, the placement of intracranial electrodes was successfully guided by FDG-PET findings, demonstrating a good concordance between focal hypometabolism, epileptogenic zone (EZ), and extent of TTFCD. In the case of focal hypometabolism limited to a single gyrus (11 cases, 52%), the EZ was confined to this area and colocalized with the TTFCD, which was located in the deep part of the sulcus. In the case of regional hypometabolism involving several gyri or lobes (10 cases), the EZ was included within the hypometabolic areas, corresponding to the maximal hypometabolic areas except in two cases in whom seizure onset was found in the orbitofrontal cortex, whereas hypometabolic areas predominated in the temporal pole. These data have been fully described in a previous study (Chassoux et al., 2010). Among the 16 MRI-positive patients investigated by SEEG, a good concordance between MRI features, EZ, hypometabolism, and TTFCD extent was found in nine cases (56%). In most of these cases, MRI abnormalities were confined to a single gyrus located near the convexity of the brain. In four other patients in whom hypometabolic areas were more extensive than the EZ and the MRI lesion, the EZ corresponded to the areas of maximal hypometabolism. In the three remaining cases, the EZ was more extensive than the MRI lesion but was discordant with the PET data. The misleading PET findings mainly corresponded to an orbitofrontal or insular TTFCD location. For these seven patients in whom the EZ did not correspond to imaging findings, PET was helpful in more than half. These results suggest that when MRI and PET findings are strictly concordant and involve a single gyrus, performing a resection based on imaging data may be a rational strategy. Conversely, when hypometabolic areas are detected beyond the MRI abnormalities, invasive monitoring remains mandatory, especially in orbitofrontal or insular regions.
Surgical data and neuropathologic findings
TTFCD location was extratemporal in all patients but one, who was in the MRI-positive group. Involvement of frontal and functional areas (frontal posterior premotor areas, precentral, central or postcentral regions, and insular or Broca areas) was characteristic and did not differ significantly between the two groups. The more limited resections were performed in the MRI-negative group, and were confined to a single gyrus in more than half of the cases. Five patients in the MRI-negative group underwent a second procedure after incomplete seizure control following the first operation, including stereotactic thermocoagulation (Guénot et al., 2004) in two cases. All of these five patients had SEEG prior to surgery. In the MRI-positive group, larger resections including the MRI lesion and the adjacent gyri were mostly carried out, with more restricted resections being performed in functional areas. Early postoperative status did not differ between the two groups. A motor deficit was frequent (45%) after resections in premotor, precentral, central, or postcentral areas. A total recovery was mainly observed in the MRI-negative group, in which cortical resections were the most limited, whereas a permanent mild disability (such as slowness in finger tapping) was found only in the MRI-positive group. Seizure outcome was similarly favorable in both groups. Engel class I was obtained in 22 MRI-negative patients (88%) and 35 MRI-positive patients (94%). In the MRI-negative group, seizure-free outcome was obtained in three of the five patients who underwent a second procedure, including the two patients treated by stereotactic thermocoagulation. The four patients who were operated without SEEG were also seizure free. Although class IA was more frequent in MRI-positive patients (75%) than in MRI-negative patients (56%), the difference was not statistically significant. Histologic findings demonstrated typical features of TTFCD, with balloon cells in 20 MRI-negative cases (80%) and 36 MRI-positive cases (97%), mostly involving a single gyrus in both groups. Large TTFCDs, extending to several gyri, were predominantly found in the MRI-positive group, whereas the smallest TTFCDs, corresponding to a very small number of giant dysmorphic neurons located at the bottom of a sulcus were found in 9 MRI-negative patients (36%).
We report a series of 62 patients who underwent surgery for severe intractable epilepsy due to TTFCD including a large proportion of MRI-negative cases (40%), allowing us to make an extensive phenotypic comparison with MRI-positive patients. We found that the electroclinical features of MRI-negative patients did not differ from those of MRI-positive patients, except for SRE, which was significantly more frequent (p = 0.028) in the MRI-negative group. Early onset of severe partial epilepsy with high seizure frequency, stereotyped seizures, extratemporal location (and especially in the frontal lobe), and involvement of eloquent cortex was observed in the whole population. Despite the negative MRI findings, a seizure-free outcome was obtained in 88% of patients in the MRI-negative group, a similar proportion to that obtained in the MRI-positive group. These results suggest that patients with cryptogenic sleep-related partial epilepsy, whether children or adults, should be actively considered for surgery.
The strong SRE predominance in the MRI-negative group was unexpected. Because nearly 60% of our patients had frontal lobe epilepsy, we assumed that the high proportion of nocturnal and sleep-related seizures that we found in this series was due to this epilepsy localization. However, in the MRI-negative group, the presence of SRE was unrelated to the TTFCD location. Interestingly, a recent study pointed out the role of TTFCD as an independent risk factor for SRE (Nobili et al., 2009). The authors postulated that the peculiar activity displayed by the dysplastic cortex, namely the rhythmic spikes or polyspikes and bursts of fast discharges interrupted by suppression of activity, tending to recur pseudoperiodically during sleep, and may limit or reduce ictal events during wakefulness, whereas spread to surrounding cortex could be facilitated during sleep. This hypothesis may account for the high incidence of SRE in MRI-negative cases, in which dysplastic neurons that correspond to the lowest critical mass and recruitment of nondysplastic cortex during sleep is potentially required for seizure generation, whereas, in large TTFCDs containing a higher mass of dysplastic neurons, seizures may occur as readily during the waking state as during sleep. These results support the previous observations (Nobili et al., 2009) and shed light on the relationship between TTFCD and SRE.
Although partial epilepsy was reported in all previous studies on TTFCD, no specific findings were highlighted. We stress the impressive stereotyped characteristics of seizures regardless of their high frequency, pointing to a well-localized brain area. In our experience, a careful analysis of ictal semiology is indeed crucial for TTFCD localization and is particularly helpful when MRI is negative. In addition, we note that secondary generalized seizures were exceptional, although some bilateral motor manifestations could occur in central seizures.
The question of cognitive impairment remains controversial. We previously reported that mental retardation was mostly observed in children with a large dysplastic lesion and early onset of epilepsy, whereas normal functioning or mild impairment was seen in adults with small TTFCD (Chassoux et al., 2000). In the present series, cognitive impairment was relatively rare, with a low IQ (<80) in no more than 16% of adults or children, in the MRI-negative or MRI-positive groups alike. Mental retardation was found only in a few children with positive MRI. Other series have reported a wide range of cognitive impairment rates, ranging from 35–68% (Lawson et al., 2005; Widdess-Walsh et al., 2005; Krsek et al., 2008). Contradictory data according to FCD subtypes have been reported, mental retardation being more frequently observed in TTFCD than in mild FCD in some studies (Tassi et al., 2002; Widdess-Walsh et al., 2005), whereas the opposite has been reported by others (Krsek et al., 2008). Furthermore, a more severe impairment in FCD type IIA (without balloon cells) than in IIB (with balloon cells) has been described (Lawson et al., 2005), a finding that was not confirmed by others (Widdess-Walsh et al., 2005; Krsek et al., 2008). Difficulty in assessing the FCD subtype by histology may explain this discrepancy (Sisodiya et al., 2009) but the heterogeneity of cognitive profiles associated with TTFCD provides evidence of a multifactorial origin.
Localized interictal and ictal EEG was reported in 40–70% of the cases (Lerner et al., 2009) but neurophysiologic data are mostly considered nonspecific. A striking pattern of continuous or subcontinuous rhythmic spikes or sharp waves previously described in about 50% of TTFCD cases (Palmini et al., 1995; Gambardella et al., 1996; Raymond & Fish, 1996; Chassoux et al., 2000) was no longer considered characteristic in a recent series (Lerner et al., 2009). We found such pseudorhythmic activity in 44% of MRI-negative cases, the same proportion as in MRI-positive cases, and consider that such activity is highly suggestive of TTFCD, a key point when investigating a cryptogenic epilepsy patient. Furthermore, intralesional interictal activity and ictal discharges recorded during SEEG are strikingly similar in MRI-negative and MRI-positive cases, providing an electrical “signature” of the dysplastic tissue. Of note, this peculiar activity may be spatially very limited and recorded in only one or two contacts of a single intracranial electrode. This may limit the detection of very small TTFCD, which can be missed if the electrode is not exactly placed within the lesion. Therefore, the contribution of other imaging techniques (especially FDG-PET coregistered with MRI) for TTFCD localization is mandatory (Chassoux et al., 2010). In addition, when a single hypometabolic gyrus is detected, a successful focal resection may be performed without invasive monitoring, as was the case for four patients in the current series. We suggest criteria for operating without SEEG when MRI is negative as follow: (1) clinical and EEG features highly suggestive of TTFCD (including sleep-related epilepsy); (2) strongly localized hypometabolism corresponding to a single gyrus that can be totally removed without functional risk; and (3) frontal location (except in eloquent cortex).
In our experience, the usefulness of PET is not limited to MRI-negative cases; indeed, it contributed to the delineation of the dysplastic cortex in 65% of the MRI-positive cases. In the latter cases, we found that the SEEG dysplastic “core” corresponded either to the lesional area as visible on MRI when PET and MRI were strictly concordant, or to strongly hypometabolic cortical areas when PET abnormalities were more extensive than the MRI lesion. These correlations were robust and reproducible, allowing us to perform accurate and restricted resections, particularly in functional areas. In addition, coregistration of PET with MRI, allowed accurate delineation of the TTFCD in two-thirds of cases. This finding avoids the use of invasive monitoring in an increasing number of MRI-positive patients. However, the information provided by PET was inferior to that provided by MRI in patients with a very high seizure frequency, especially if seizures occurred during the PET examination. In addition, misleading metabolic data were observed in some TTFCD locations, principally in orbitofrontal and insular regions, in which temporal hypometabolism was predominant, likely due to the strong connections between these areas and early spreading of ictal discharges. Therefore, SEEG remains indicated in some selected MRI-positive cases.
We propose that cryptogenic SRE of early onset and extratemporal location are highly suggestive of TTFCD and justify pursuing the presurgical workup by further investigations, especially FDG-PET and, if necessary, invasive monitoring.
The authors thank Dr Hélène Mellerio who performed the statistical analysis and Prof. Olivier Dulac for reviewing the manuscript. They also thank Drs Frédéric Beuvon, Sonia Baudoin-Chial, Sebastian Rodrigo, Jacques Laschet, and all the SHFJ team (especially Patrick Bodilis, Dr Philippe Gervais. and Prof. Pascal Merlet) for their contribution to this study.
None of the authors has any conflict of interest to disclose.
We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
All the authors were involved in drafting and revising the article.