Address correspondence to Martin Pail, Department of Neurology, St. Anne’s University Hospital and Faculty of Medicine, Pekařská 53, 656 91 Brno, Czech Republic. E-mail: firstname.lastname@example.org
Purpose: To determine whether voxel-based morphometry (VBM) might contribute to the detection of cortical dysplasia within the temporal pole in patients with mesial temporal lobe epilepsy and hippocampal sclerosis (MTLE/HS).
Methods: Eighteen patients with intractable MTLE/HS and 30 sex- and age-matched healthy controls were included in the study. All of the patients fulfilled the diagnostic criteria for MTLE/HS and underwent anteromedial temporal resection. VBM without a modulation step was applied to the magnetic resonance (MR) images of the brain. Statistical parametric maps were used to compare structural characteristics such as gray matter concentration (GMC) within the temporal pole among patients and controls separately. The acquired data were then statistically analyzed to determine the congruency between visually inspected MR imaging (MRI) scans and VBM results in the detection of morphologic abnormalities in the temporal pole compared to postoperative histopathologic findings of cortical dysplasia.
Key Findings: Histopathologic examination revealed cortical dysplasia within the temporal pole in 11 patients. In detail, according to Palmini’s classification, mild malformations of cortical development (mMCDs) were disclosed in three patients, focal cortical dysplasia (FCD) type Ia in three patients, and FCD type Ib in five patients. Some type of structural temporal pole abnormality was suggested by VBM in 14 patients and by visually inspected MRI scans in 11 patients. The results of VBM were in agreement with the presence/absence of cortical dysplasia in 13 patients (72.2%); this correspondence was significant (p = 0.047). In one case, VBM was false negative and in four cases it was false positive. There was congruence between the results of visual analysis and histologic proof in 55.6% of examined patients, which was not significant.
Significance: We found that VBM made a superior contribution to the detection of temporopolar structural malformations (cortical dysplasia) compared to visual inspection. The agreement with postoperative histopathologic proof was clearly significant for VBM results and nonsignificant for visual inspection.
Mesial temporal lobe epilepsy (MTLE) is one of the most frequent forms of localization-related epilepsy and the most common human intractable epilepsy. Hippocampal sclerosis is the underlying pathology in the majority of cases of MTLE, and is histopathologically characterized by gliosis and neuronal loss. Recent studies in patients with mesial temporal lobe epilepsy and hippocampal sclerosis (MTLE/HS) repeatedly revealed structural and functional abnormalities in regions outside the affected hippocampus. Changes in tissue volume or concentration (both gray and white matter) as well as functional changes (disclosed in electroencephalography [EEG], magnetic resonance spectroscopy [MRS], positron emission tomography [PET], single photon emission computed tomography [SPECT] and subtraction ictal SPECT coregistered to MRI [SISCOM] studies) were repeatedly shown extending to other brain regions, mainly ipsilateral to the side of the epileptic focus (Guye et al., 2006; Nelissen et al., 2006; Fojtiková et al., 2007; Keller & Roberts, 2008; Wehner & Lüders, 2008; Brázdil et al., 2009; Pail et al., 2010).
Several neuropathologic subtypes of FCD are distinguished based on histopathologic features and reflecting degree of abnormality. Palmini’s classification (2004) proposes mild malformation of cortical development (mMCD) types I and II and FCD types I and II. FCD type I refers to architectural disturbances of cortical lamination (type Ia) and cytoarchitectural abnormalities (type Ib). FCD type II is characterized by cortical dyslamination with dysmorphic neurons without (type IIa) and with balloon cells (type IIb). The revised Blümcke classification proposes a three-tiered classification system. FCD type I denotes either radial (FCD type Ia) or tangential (FCD type Ib) dyslamination of the neocortex. FCD type II resembles the Palmini classification. The principal modification to Palmini’s classification is the addition of FCD type III, which occurs in combination with hippocampal sclerosis (FCD type IIIa), adjacent to epilepsy-associated tumors (FCD type IIIb) or contiguous vascular malformations (FCD type IIIc), or is associated with any other principal lesion acquired during early life (FCD type IIId) (Palmini et al., 2004; Blümcke et al., 2009, 2011).
Magnetic resonance imaging (MRI) plays a principal role in the noninvasive presurgical evaluation of patients with intractable epilepsy and the detection of MCD. Nevertheless many patients are classified as MRI-negative because their cortical and subcortical morphologic changes are invisible, borderline, or subtle. MRI postprocessing methods have recently been established to improve the detection and localization of structural abnormalities. The most frequently used postprocessing methods are voxel-based morphometry (VBM) and diffusion tensor imaging (DTI) (Wehner & Lüders, 2008; Colombo et al., 2009). Others include junctions and automatic curvilinear models (Bastos et al., 1999; Huppertz et al., 2005). The areas defined as abnormal by VBM, but not visually detected as such could correspond to type I FCD or mMCD (Bonilha et al., 2006).
In this study, we aimed to investigate the contribution of VBM (an automatic whole-brain MRI analysis technique) in preoperative noninvasive detection of MCD in the temporal pole in patients with intractable MTLE/HS. Statistical parametric maps (SPMs) were used to identify focal regional deviations in patients’ gray matter concentration (GMC; in comparison to a control group) within the temporal lobe pole.
We studied 18 consecutive patients with MTLE/HS (11 female, seven male) with a mean age of 40.9 years (standard deviation [SD] 9.6; range 22–56 years): 12 patients with left-sided MTLE and 6 patients with right-sided MTLE (Table 1). Patients were referred to the Brno Epilepsy Center, Department of Neurology. All the patients fulfilled the diagnostic criteria for mesial temporal lobe epilepsy with hippocampal sclerosis (MTLE/HS). The diagnosis was made according the International League Against Epilepsy (ILAE) criteria (Commission on Classification and Terminology of the International League Against Epilepsy, 1989). All of the patients had been routinely investigated, including long-term semiinvasive video-EEG (Alien Technik, Hronov, Czech Republic) monitoring (using sphenoidal electrodes), high-resolution MRI (Siemens Symphony, Erlangen Germany), and neuropsychological testing. The diagnosis of unilateral MTLE in our patients was based on a concordance of history data, ictal and interictal EEG findings, ictal semiology, neuropsychology, and neuroimaging findings. Visual inspections of the MRI scans (the MRI protocol included T1, T2, and fluid-attenuated inversion recovery [FLAIR] sequences—axial and coronal slices) by two independent physicians (radiologist and epileptologist) revealed unequivocal unilateral hippocampal sclerosis in 12 patients on the left side and in 6 patients on the right side. All of our patients had MRI evidence of unilateral hippocampal sclerosis concordant with the EEG lateralization of the epileptogenic zone. None of our patients revealed other brain structural lesions on MRI scans (with the exception of features of MCD in temporal pole; see below); patients with other types of malformations such as posttraumatic lesions, tumors, and vascular lesions were excluded from our study. All of the patients fulfilled the diagnostic criteria for MTLE/HS and underwent anteromedial temporal resection. None of the patients had undergone previous intracranial surgery; one patient underwent the implantation of a vagus nerve stimulation system (Cyberonics, Houston, TX, U.S.A.) before anteromedial temporal resection. All patients had been seizure free for ≥24 h before the MRI investigation. Many studies have proven the significant variability of GMC according to gender and age (Good et al., 2001a,b; Barnes et al., 2010). To minimize the influence of these factors on our results, we used several control groups. Each consisted of 30 healthy subjects and was age and gender matched to individual patients (for details, see Table 2). The majority of the healthy subjects in the control groups were volunteers from the professional sector; no history of neurologic or psychiatric disease was presented in any controls.
Written informed consent was obtained from each participant after all of the procedures were fully explained. The study received the approval of the local ethics committee.
MR image acquisition
MR examinations were performed on a 1.5 T scanner (Siemens Magnetom Symphony, Erlangen, Germany) using a multichannel head coil. The MRI protocol for VBM included three-dimensional T1-weighted magnetization prepared rapid gradient echo (MPRAGE) sequence with time to repeat (TR) = 1.7 s, time to echo (TE) = 3.93 ms, inversion time (TI) = 1.1 s, flip angle (FA) = 15 degrees, 512 × 512 matrix size, field of view (FOV) 246 × 246 mm, and 160 sagittal slices with slice thickness = 1.17 mm.
Formalin-fixed paraffin-embedded tissues of temporal lobe resection specimens and hippocampi were available from all 18 patients. The paraffin-embedded tissue specimens, slides, and histopathology reports were retrieved from the files of the First Department of Pathological Anatomy of St. Anne’s University Hospital and independently reevaluated by two histopathologists (MH, BS). Discrepancies were resolved by consensus. All examined resected tissues were identically treated, fixed in 10% neutral buffered formalin, grossly inspected, and measured. Temporal lobe resection specimens were cut so as to obtain representative tissue slices perpendicular to the cortical surface. Hippocampal specimens were fully processed and serially sectioned perpendicular to the long axis.
Tissue slices were routinely processed and paraffin embedded. Tissue sections (5 μm) were stained with hematoxylin and eosin, evaluated under light microscopy, and reported.
The classification system reported by Palmini et al. (2004) was used to evaluate the temporal lobe resection specimens. In addition, the new classification according to Blümcke et al. (2011) was completed; see Table 3.
Table 3. Results of different diagnostic approaches (detection of structural cortical abnormality in the temporal pole)
Anatomic MRI data were analyzed using SPM8 (Wellcome Department of Cognitive Neurology, http://www.fil.ion.ucl.ac.uk) with its internal toolbox DARTEL running in MATLAB 7.5 (The MathWorks, Natick, MA, U.S.A.). Data of all the participants were segmented into gray matter images. For each control group separately, the 30 subjects were registered together and the gender- and age-specific gray matter image template was created using the DARTEL toolbox. Each patient’s gray matter image was then registered to the appropriate template. Finally, the registered images were resampled to the spatial resolution of 1.5 × 1.5 × 1.5 mm and smoothed with an 8-mm Gaussian kernel (Salmond et al., 2002).
The unmodulated and smoothed gray matter images were analyzed using analysis of variance (ANOVA) models to determine the areas of increased or decreased GMC of an individual patient compared to the respective control group. Age was included in the design matrices as a confound covariate to reduce the nuisance variability in the data. The threshold for proportional masking was set to 0.2. The analysis was restricted to the temporal lobes and mesiotemporal areas. The control-group specific masks were created manually using the control-group specific gray matter image template. The group difference was evaluated using two-sample t-test. Although the patient group contains only a single subject, and the design is thus strongly unbalanced, this approach was suggested previously (Crawford, 1998; Mühlau et al., 2009). For each patient, the contrast of patient versus respective control group resulted in the t-statistic map, which was thresholded at a p-value of <0.001 uncorrected (Fig. 1). The result was positive when there was a cluster with more than 10 suprathresholded voxels (34 mm3). Postoperative MRI scans were used to confirm that the GMC abnormalities detected by VBM were located precisely in the resected area of the temporal pole.
All brain native MR images of patients (in particular the mesiotemporal structures and the temporal pole) were analyzed independently by two separate investigators blinded to histopathologic results (IT, RK); if any discrepancies occurred, a third investigator was used (MB). Finally the results of different diagnostic approaches were compared and statistically analyzed using the attribute-agreement analysis.
The main demographic and clinical characteristics of all 18 patients are shown in Table 1. The majority of patients had no history of precipitating events. Three patients with MTLE/HS had concussion, perinatal encephalopathy, or poliomyelitis before the first seizure. Five patients had a history of febrile convulsions. The majority of patients had a history of complex partial seizures, in 11 patients with sporadic ictal seizure generalization (generalized tonic–clonic seizure, GTCS). The mean age at seizure onset was 11.2 (SD 10.7, range 0.5–43 years) (Table 1).
The results of MCD detection by different diagnostic approaches are shown in Tables 3 and 4.
Table 4. Results of FCD and HS detection by VBM (detection of structural abnormality in the temporal lobe)
Hippocampal sclerosis was proved by postoperative histopathologic examination in all patients. MCD was revealed by pathologic examination in the temporopolar resection specimens in 11 patients. In detail, according to Palmini’s classification, mMCDs were disclosed in three patients, FCD type Ia in three patients, and FCD type Ib in five patients (Table 3).
Visual evaluation of MRI scans
Unilateral atrophy of the hippocampus was visually determined in all patients; the side was coincident with the side detected first by EEG and subsequently histopathologically. Visual analysis of MR scans suggested signs of MCD in the temporal pole in 11 cases (four false negatives and four false positives); in all patients the structural changes (signs of MCD) in the temporal pole were seen on the same side that the HS were observed. This approach correctly detected presence/absence of temporopolar MCD in 10 patients (55.6%). Statistical analysis did not detect a significant correspondence between visually analyzed results and histopathologic findings of MCD in TP (Z = 0.275 and p = 0.39).
VBM analysis of MRI scans
In the group of patients with MTLE/HS, a GMC abnormality in the temporal pole was detected by VBM in 14 cases; in four patients the changes were bilateral. The abnormalities in temporal pole (on the side of MTLE) revealed by VBM were seen just in the resected area of the temporal pole compared with postoperative MRI scans. The results of VBM were in congruence with the presence/absence of cortical dysplasia (confirmed by postoperative histopathologic proof) in 13 patients (72.2%); this correspondence was significant (Cohen kappa = 0.37; Z = 1.680 and p = 0.047). VBM was false negative in one patient and false positive in four patients. In most cases, VBM revealed MCD in patients by increased GMC. However, in three patients FCD was confirmed by decreased GMC. HS was revealed by VBM in 12 cases (vice versa in most patients with decreased GMC), in 9 patients just unilaterally, and the side was coincident with the side detected visually and histopathologically. Structural abnormality was also revealed by VBM in the contralateral hippocampus in three patients. For more detailed information, see Table 4.
MRI plays a principal role in the noninvasive presurgical evaluation of patients with intractable MTLE. MRI enables the detection of HS and eventually of coexistent morphologic alterations in cortical development, and contributes to the selection of suitable subjects for surgical treatment. Although all of our patients had visually confirmed MRI evidence of unilateral hippocampal sclerosis concordant with the EEG lateralization of the epileptogenic zone, VBM disclosed HS in only 12 cases. We suppose for the detection of tissue atrophy (in the hippocampus) VBM with its modulation step is better, as it has higher sensitivity to gray matter volume (Brázdil et al., 2009; Pail et al., 2010). Nevertheless, as can be seen in our results, many patients are incorrectly categorized as MRI negative (MCD in the temporal pole). Visual detection correctly suggested some structural abnormality that corresponded with the presence/absence of MCD in the temporal pole entirely in 55.6% of 18 patients. This result is deficient within presurgical evaluation and prohibits sufficient surgical intervention. Identification of FCD can be difficult due to subtle MRI changes; nevertheless, a majority of these lesions can be visually detected using FLAIR sequences (Rajan et al., 2009). High-quality MRI is required. FCD can be visually detected by means of characteristic MRI features (cortical thickening, abnormal gyral and sulcal contours, blurring of the gray matter-white matter interface, abnormal signal intensity in the cortex, subcortical white matter, and focal hypoplasia/atrophy) (Gómez-Ansón et al., 2000; Tassi et al., 2002; Colombo et al., 2003; Krsek et al., 2008; Colombo et al., 2009). The visual detection of FCD is imperfect despite the knowledge of signs indicating FCD.
To facilitate detection and localization of some subtle abnormalities, several neuroimaging approaches, methods, and modalities have been established in the presurgical evaluation of patients with intractable epilepsy. These include PET, coregistered PET/MRI, ictal and interictal SPECT to visualize metabolic alterations of the brain in the ictal and interictal states, and SISCOM (Fauser et al., 2004; Wehner & Lüders, 2008; Colombo et al., 2009).
Other methods (software) use and statistically process MRI scans to improve detection of subtle MCD. In the literature, these methods are called MRI postprocessing (such as VBM, voxel-based three-dimensional MRI, curvilinear reformatting, junctions, DTI, and partial differential equation) (Bastos et al., 1999; Kassubek et al., 2002; Colombo et al., 2009; Rajan et al., 2009). These specific strategies enable the detection of morphologic changes in the brain structure not detected by simple visual inspections of MRI scans. Because of its superior contribution in the determination of MCD, DTI in particular might become an indivisible part of the presurgical evaluation of the brain (Colombo et al., 2009).
In our study, we evaluated VBM. Two approaches are commonly used to test the patient difference from the control group: the analysis of the standardized voxel-based deviation of patient data from the control group (Bonilha et al., 2009) or the two-sample t-test. We used the latter, as this approach has been shown to be more conservative (Crawford, 1998; Mühlau et al., 2009) and thus more suitable in studies that are not sharply focused on a small specific region. This approach may result in some false-positive results. The VBM correctly and in a statistically significant manner disclosed the presence/absence of structural abnormality in the temporal pole (MCD) in 72.2% of patients (p = 0.047). In one patient VBM was false negative and in four patients VBM was false positive. The explanation of the high false positivity of VBM might be due to diffuse GMC changes resulting from morphologic abnormalities (such as tissue atrophy) in the whole epileptogenic network beyond the hippocampus, as has been previously described (Keller & Roberts, 2008). The temporal pole is the part of this network. These results might also be due to the lower specificity of VBM (a result of not correcting for multiple comparisons in our study). Of interest, VBM revealed morphologic disruption in all three patients with mMCD type I. Distinctive visual radiologic features of mMCD in MRI scans can also be seen (Krsek et al., 2008). We suppose that MCD (including mMCD) might result in some histochemical alterations in neural tissue and, along with potential dysmyelination and a reduced number of myelinated fibers, might influence the MR signal. VBM might detect these surrounding, sometimes remote, and widespread changes. For this reason, dysplasia may not correspond strictly with the area proven to be pathologic by VBM.
In most cases, VBM disclosed MCD in patients by increased GMC, as had occurred in previous studies (Kassubek et al., 2002; Bonilha et al., 2006). We confirmed the importance of testing for both increases and decreases in relative GMC when using VBM (Bruggemann et al., 2009). Although decreased GMC might reflect a more likely atrophy of the temporal pole, it appears that subtle MCD can be hidden in more widespread structural changes such as atrophy. The atrophy of the temporopolar and mesiotemporal structures was described as a common feature of MTLE/HS (Keller & Roberts, 2008). Our study also confirmed an association of HS with macroscopic or microscopic cortical dysplasia (FCD III, Blümcke, 2011), as presented in the preceding text. We revealed and confirmed VBM as a postprocessing MRI technique that may help identify additional brain pathologic conditions other than HS in patients with TLE.
Anteromedial temporal resection and selective amygdalohippocampectomy are the most frequently used approaches in epilepsy surgery of intractable MTLE/HS. Reports on the postoperative outcome of patients with HS and associated microscopic cortical dysplasia provide conflicting evidence. Early studies presented a higher risk for seizure recurrence after epilepsy surgery (although both pathologies are removed) in contrast with patients who had MTLE with hippocampal sclerosis only (Palmini et al., 1994). Recent studies revealed favorable seizure outcomes in both groups of patients (Fauser et al., 2004, 2008; Morales Chacón et al., 2009, Tassi et al., 2009). Of interest, three fourths of the patients in whom bilateral VBM changes in the temporal pole were disclosed had a poor outcome (Engel class IIIA or IV) (Engel et al., 1993). This result seems to strongly indicate that some type of malformation in cortical development might also be in the contralateral temporal pole in these patients. The outcome is also influenced by the type of FCD. Our results show that the patients with FCD Ib had a worse outcome, as defined by Engel, than the patients with FCD Ia. These observations might also be due to the incomplete resection of the dysplastic tissue (Bonilha et al., 2006). Nevertheless the patients with bilateral hippocampal atrophy according to VBM did not have an inferior outcome. However, VBM, with its modulation step, is better for the detection of HS.
Less invasive surgical interventions, such as stereotactic radiofrequency, were recently established for the treatment of intractable TLE. This method provides selective surgical intervention of the medial structures (hippocampus, amygdala, and parahippocampal gyrus), while minimizing neocortical resection (especially temporal pole) and postsurgical complication (Talairach et al., 1974; Kalina et al., 2007). Several studies revealed better neuropsychological outcome after limited resections compared with anteromedial temporal resection (Clusmann et al., 2002; Lacruz et al., 2004). The temporal pole is supposed to be part of the limbic system and plays a pivotal role in social-emotional processes (Kotašková & Marusič, 2010). It was shown that patients after successful resection (anteromedial temporal resection) experience social maladaptation (Olson et al., 2007). The role of neocortical structures, and especially of the temporal pole, in the genesis of temporal lobe seizures has been discussed intensively and repeatedly. Extensive invasive video-EEG study using depth electrodes showed that the temporal pole is involved at the onset of seizures in about one third of patients with MTLE/HS (Chabardès et al., 2005). In an animal study, Germano et al., (1996) presented HS as a subsequent pathology resulting from increased susceptibility to febrile seizures (during which hippocampus sustains damage) caused by neuronal changes (FCD) in the temporal lobe. This explanation is also presented in a recent classification of FCDs (Blümcke et al., 2011). The mechanism of epileptogenicity and the relevance of a dual pathology or FCD IIIa are uncertain (Gómez-Ansón et al., 2000; Fauser & Schulze-Bonhage, 2006; Blümcke et al., 2011).
Limited resection (e.g., selective amygdalohippocampectomy, preserving the temporal pole) is possible (indicated) in patients with HS only. If a common pathology (HS and MCD in temporal pole) is presented, a favorable postsurgical outcome might be expected providing that both lesions are removed (Marusic et al., 2007). There is no presurgical noninvasive diagnostic method that may reliably detect MCD beyond HS and enable the selection of patients who might undergo (benefit from) limited resection or stereotactic radiofrequency intervention. MCD may not overlap the seizure onset zone, and the exact determination of seizure onset zone is only possible using invasive EEG monitoring (Guerrini & Barba, 2010).
Our study suggests that VBM might be helpful in the detection of subtle MRI abnormalities (cortical dysplasia) in the temporal pole. The agreement with postoperative histopathologic proof was clearly significant for VBM results and nonsignificant for visual inspection. The research shows that VBM is a more beneficial method than visual evaluation of MRI scans. If some subtle pathology in patients with MTLE is detected in the temporal pole by VBM, limited surgical procedures should be carefully considered. The limited number of patients included in this study certainly has to be taken into account, and further confirmation of our findings should be explored.
We thank Anne Johnson for grammatical assistance. The study was supported by the project “CEITEC – Central European Institute of Technology” (CZ.1.05/1.1.00/02.0068) from European Regional Development Fund and by MŠMT ČR Research Program no. MSM0021622404.
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. None of the authors has any conflict of interest to disclose.