Asymmetric cortical surface area and morphology changes in mesial temporal lobe epilepsy with hippocampal sclerosis

Authors


Address correspondence to Dr. Gianpiero Cavalleri, The Department of Molecular and Cellular Therapeutics, The Royal College of Surgeons in Ireland, Dublin 2, Ireland. E-mail: gcavalleri@rcsi.ie

Summary

Purpose:  To date, magnetic resonance imaging (MRI)–based studies of the cerebral cortex in mesial temporal lobe epilepsy (MTLE) have focused primarily on investigating cortical volume and thickness. However, volume is a composite of surface area and thickness, each reflecting distinct neurobiologic and genetic processes. The goal of this study was to investigate cerebral cortex (1) surface area, (2) surface geometric distortion, and (3) thickness in MTLE with hippocampal sclerosis (HS).

Methods:  Seventy patients with “sporadic” unilateral MTLE + HS and 40 healthy controls underwent T1-weighted MRI. Processing MR images using an automated cortical surface reconstruction method (FreeSurfer), we quantified cortical surface area, surface geometric distortion (metric distortion), and thickness at each vertex across the entire cortex. Differences between patients and controls were determined using generalized linear models. Separate linear regression models were employed to assess the relationship between cortical surface area and hippocampal volume as well as a series of important clinical features of the condition.

Key Findings:  We detected an asymmetric reduction in cortical surface area, predominantly in ipsilateral mesial and anterior temporal lobe subregions, of patients with MTLE + HS. Changes in surface geometric features were also evident and closely mirrored surface area patterns. In contrast, cortical thinning appeared dispersed across the cortex bilaterally. The regression models revealed that ipsilateral hippocampal volume was a significant predictor of temporal lobe surface area changes.

Significance:  Our findings indicate that contraction in surface area, rather than cortical thinning, explains ipsilateral mesial and anterior temporal lobe atrophy in patients with MTLE with HS. Furthermore, the alterations in surface geometry indicate folding abnormality involving the same regions. Cortical surface changes may represent sequelae of the disease or deviant cortical development.

Mesial temporal lobe epilepsy (MTLE) is the most prevalent form of medically intractable epilepsy. The pathologic hallmark of MTLE is hippocampal sclerosis (HS), which is found in approximately 65% of patients (Wieser and ILAE Commission on Neurosurgery of epilepsy, 2004). HS is characterized histologically by cellular loss and synaptic reorganization in hippocampal subregions, and often can be detected using magnetic resonance imaging (MRI) through the identification of hippocampal atrophy and MR signal abnormalities (Berkovic et al., 1991). Hippocampal atrophy is often accompanied by subtle brain structural changes in other subcortical regions (Bernasconi et al., 2003; McDonald et al., 2008b), the cerebellum (Hermann et al., 2005), and several neocortical regions (Doherty et al., 2003; Scanlon et al., 2011). The underlying driver of these structural changes remains poorly understood but is likely the product of environmental (e.g., initial precipitating insults and seizure-induced damage) and polygenic factors (Briellmann et al., 2002; Mathern et al., 2002; Labate et al., 2011).

Until recently, MRI-based studies in MTLE have focused primarily on identifying disease-related cerebral cortical volume abnormalities (Keller & Roberts, 2008). Using voxel-based morphometry (VBM) and region-of-interest (ROI) volumetric analyses, reduction in cortical gray matter volume (GMV) has been reported for several limbic and neocortical regions predominantly ipsilateral to the side of seizure focus (Keller & Roberts, 2008). However, cortical volume is a composite of surface area and thickness, and recent work has illustrated that these two measurements are driven by apparently distinct evolutionary, cellular, and genetic mechanisms (Im et al., 2008; Panizzon et al., 2009). Therefore, the study of cortical surface area and thickness informs on distinct neurobiologic and genetic processes that could contribute to the pathophysiology of neurologic conditions such as MTLE.

Various studies have explored cortical thickness in MTLE and reported localized changes dispersed bilaterally across the cortex (Lin et al., 2007; McDonald et al., 2008a; Mueller et al., 2009; Labate et al., 2011). Bilateral thinning of up to 30% in regions localized within the frontal, lateral temporal, and occipital lobes were reported by Lin et al. (2007). Subsequent cross-sectional studies, in comparable population sizes, reported similar findings in patients with mild and refractory MTLE. Cortical thinning appeared progressive in pharmacoresistant MTLE, and this has been attributed to seizure-induced neuronal damage involving hippocampal projection pathways (Bernhardt et al., 2009).

Previous studies of cortical surface geometric features in MTLE included assessments of cortical folding complexity and gyrification. The findings of these studies have been inconsistent, with some authors suggesting a global increase in gyrification (Oyegbile et al., 2004; Ronan et al., 2007) and others reporting decreased cortical complexity (Lin et al., 2007). More recently, Voets et al. reported evidence of increased temporolimbic surface folding complexity, associated with increased hippocampal malrotation in MTLE + HS. The authors interpreted the findings as evidence of preexisting neurodevelopmental abnormality (Voets et al., 2011). The measures of folding complexity and gyrification can be affected by both a localized change in the surface area or angular and rotational displacement related to the development of gyri and sulci. Previous studies have not separated these two variables.

In the present study, we aimed to investigate cerebral cortex (1) surface area, (2) surface geometric distortion, and (3) thickness in patients with “sporadic” unilateral MTLE + HS. Surface geometric distortion was quantified by calculating the Jacobian determinant of deformation, which measures the degree of metric distortion required to register each subject’s cortical surface to an average template (Fischl et al., 1999a,b; Wisco et al., 2007).

Methods

Study participants

Patients

Patients with a clinical diagnosis of MTLE + HS were recruited from Beaumont Hospital and St. James’s Hospital, both tertiary epilepsy centers in Dublin, Ireland. All patients underwent a comprehensive review that confirmed the clinical diagnosis of MTLE + HS. The side of seizure activity focus was determined by a comprehensive evaluation including a combination of seizure semiology, ictal/interictal electroencephalography (EEG), video-telemetry recordings, and qualitative inspection of MRI films for evidence of HS. Patients with evidence of any lesion other than HS were excluded. In total, 70 patients (30 male and 40 female) were included. Side of seizure focus (and HS) was left in 35 patients (16 male) and right in 35 patients (14 male). Unilateral hippocampal atrophy was confirmed in all cases by identifying volume loss beyond 2 standard deviations (SD) of the mean of healthy controls (n = 40). All patients had “sporadic” disease and reported no history of epilepsy or febrile seizures in first- or second-degree relatives. Sixty-one of our patients (87.1%) were classified as refractory to medical treatment as defined by Kwan et al. (2010).

Controls

Our control population was chosen from a bank of control MR image data at Beaumont Hospital and consisted of 40 healthy individuals (18 male and 22 female), demographically matched to our patient groups; see Table 1 for additional details of study participants.

Table 1.   Demographic characteristics of study participants
 Healthy controlsLeft MTLE + HSRight MTLE + HS
  1. MTLE + HS, mesial temporal lobe epilepsy with hippocampal sclerosis; SD, standard deviation; IPIs, initial precipitating insults. None of the healthy controls reported any history of febrile seizures or intracranial infection.

Number403535
Age: mean (SD)33.7 (9.9)36.8 (11.4)36.6 (10.6)
Gender: number (%)   
 Male18 (45)16 (45.7)14 (40)
 Female22 (55)19 (54.3)21 (60)
Age at onset: mean (SD)14.2 (11.3)12.6 (10.5)
Epilepsy duration: mean (SD)22.4 (14.4)23.1 (13.6)
IPIs: number (%)   
 Febrile seizures17 (48.6)16 (45.7)
 Intracranial Infection1 (2.8)3 (8.5)
 Status epilepticus4 (11.4)5 (14.2)

The research ethics committees of both hospitals independently approved this study and written informed consent was obtained from all participants.

MR image acquisition

MRI scans of the brain were acquired for all participants using 1.5 Tesla MRI scanner (Signa, GE, Milwaukee, WI, U.S.A.) at Beaumont Hospital. A three-dimensional (3D) T1-weighted spoiled gradient recalled sequence (SPGR; TR/TE = 10.1/4.2 msec, msec, TI = 450 msec, flip angle = 20 degrees, field of view (FOV) = 24 × 24 cm2, matrix = 256 × 256) with 124 sagittal slices (slice thickness = 1.5 mm) was used to acquire the images for all participants.

MR image processing

MR images were transferred in DICOM format to a dedicated Linux workstation (Ubuntu version 7.04). All images were processed using FreeSurfer, a fully automated image analysis package (version 4.50, https://surfer.nmr.mgh.harvard.edu). FreeSurfer was used for cortical surface reconstruction and segmentation of cortical and subcortical structures. The FreeSurfer process has been described in detail previously (Dale et al., 1999; Fischl et al., 1999a,b) and has undergone extensive investigations to assess its accuracy, validity, and applicability (Han et al., 2006). In summary, the process started with removal of nonbrain tissue, intensity normalization, transformation to Talairach space, and segmentation of the subcortical white matter and deep gray matter structures. The gray-white matter boundary was then determined and any topologic defects were corrected. This is followed by deformation of gray-white matter boundary outward to generate the pial surface as indicated by intensity gradient (Dale et al., 1999). Once generated, convoluted reconstruction of the cortical surfaces was inflated to a smooth surface, allowing both sulcal and gyral folds to be viewed. Subsequently, each vertex on the inflated surface was registered to a sphere (Fischl et al., 1999a), which was then mapped to an average surface template for optimal alignment of the cortical fold patterns in a way that minimizes metric distortion (Fischl et al., 1999b). Careful visual quality checks of all surfaces were performed at several stages of FreeSurfer’s processing stream and any geometric inaccuracies were manually corrected as recommended by the software guidelines.

MR image analysis

Cortical surface area

The area assigned to each vertex on the gray-white matter surface was calculated as the average area of all triangles that surround the vertex. The size of each triangle is constant; however, the number of triangles varies according to brain size. To compare patients to controls and obtain maps of surface area alterations, we applied the method used by Joyner et al. (2009); Palaniyappan and Liddle (2012) and Palaniyappan et al. (2011). This method requires deformation of individual surfaces using a spherical atlas registration procedure followed by registration of individual spheres into the common coordinate system. This results in a standard number of tessellations across each individual’s cortical surface with surface area values assigned to each vertex redistributed to reflect the relative expansion and contraction of the cortical sheet around each vertex. The constant number of vertices generated for each subject enables group-wise quantification of cortical surface area changes across the entire cortical mantle.

Surface geometric distortion

Cortical surface geometric distortion was quantified by measuring the degree of metric distortion (Jacobian determinate of deformation) required to register cortical folding patterns in each subject to an average template at each vertex across the entire cortex (Fischl et al., 1999a). This measure captures the linear, rotational, or angular distortion in the cortical mantle (Wisco et al., 2007).

Cortical thickness

At each vertex, the cortical thickness was calculated as the average of the shortest distance between the gray–white matter surface and the gray matter–CSF (pial) surface.

Intracranial and hippocampal volume

An estimate of intracranial volume was provided by FreeSurfer based on the transformation of each subject’s brain into Talairach space (Buckner et al., 2004). Hippocampal volume was quantified using FreeSurfer’s segmentation process.

Clinical characteristics

Clinical information was collected from patients and their relatives using structured interviews at the time of scanning as described in Alhusaini et al. (2012). Interviews were supplemented with information collected from medical records when necessary. The following clinical variables were collected: age at seizure onset, disease duration, history of initial precipitating insults (IPIs), and estimates of the lifetime number of partial seizures and partial-onset seizures that evolved to generalized convulsive events (i.e., secondary generalized tonic–clonic seizures, SGTCS). The following were considered as IPIs: febrile seizures (FS), intracranial infection(s), and status epilepticus. Partial seizures and SGTCS were defined according to the International League Against Epilepsy (ILAE) guidelines (Commission of Classification and Terminology of the International League Against Epilepsy, 1981). The number of partial seizures and SGTCS were estimated as described previously (Alhusaini et al., 2012).

Statistical analysis

Demographic variables

Age and gender were compared between patient and control groups using t and chi-square tests. The Mann-Whitney U-test was used to compare left and right MTLE + HS patient groups with respect to disease duration, age at seizure onset, and frequency of seizures. In addition, frequency of IPIs in the left compared with the right MTLE + HS patient group was assessed through a chi-square test.

MRI data analysis

We analyzed differences in cortical surface area, surface geometric distortion (metric distortion), and thickness by computing a general linear model (GLM) of the effect of case–control status on each measure across the entire cortex as implemented in the Query Design Estimate Contrast (QDEC) interface of FreeSurfer. Cortical surface area, metric distortion, and thickness differences were tested between (1) left MTLE + HS patients and controls and (2) right MTLE + HS patients and controls. As regional surface area of the cerebral cortex had been shown to correlate highly with head size, estimated intracranial volume (ICV) and age were included as covariates in surface area and metric distortion analyses (Dickerson et al., 2009). Age and gender were included as covariates in cortical thickness analysis due to their effect on cortical thickness (Barnes et al., 2010).

Correlation with ipsilateral hippocampal volume and clinical features

Here we focused on cortical surface area. A GLM model, controlling for ICV and age, was employed in patients and controls separately to assess the correlation between cortical surface area at each vertex and hippocampal volume in each hemisphere. Similarly, a GLM model, controlling for the effect of ICV and age, was performed to assess the correlation between cortical surface area and disease duration, age at seizure onset, and lifetime number of partial seizures and SGTCS. To assess the influence of IPIs on cortical surface area, patients with a previous history of IPIs were compared to patients without an IPI history using a separate GLM model (covariates: ICV and age).

All data were smoothed with a 20-mm full width half maximum (FWHM) Gaussian kernel. We expected any meaningful surface changes to involve the macroscopic extent of sulci/gyri. A smoothing index of 20 mm ensures such changes are captured without a significant increase in false-positive clusters (Hagler et al., 2006). Statistical parametric maps of significant group differences in each measure and correlations were corrected for multiple comparisons using Monte Carlo permutation cluster analyses (10,000 permutations) using a cluster inclusion threshold of p < 0.05 (Holmes et al., 1996).

Results

Independent t and chi-square tests revealed no significant age or gender differences between the patients and control subjects. Mann-Whitney U-tests revealed no significant differences between the patient groups in age at seizure onset, duration of epilepsy, or frequency of seizures. Furthermore, no statistically significant difference in frequency of IPIs between the patient groups was revealed. Demographic characteristics of the cohort together with a frequency breakdown of each IPI are shown in Table 1.

Group difference in cortical surface area

Relative to controls, patients with MTLE + HS showed statistically significant regions of surface area reduction predominantly in ipsilateral temporal lobe. Figure 1 illustrates regions of surface area contraction that survived correction for multiple comparisons in left and right MTLE + HS patient groups compared to the healthy controls. In patients with left MTLE + HS, significant reduction in cortical surface area was identified in ipsilateral entorhinal cortex, parahippocampal gyrus, temporal pole, and the anterior parts of the superior, middle, and inferior temporal gyri. Smaller localized regions of surface area reduction were also identified in ipsilateral superior frontal gyrus, contralateral fusiform gyrus, and contralateral supramarginal gyrus. Patients with right MTLE + HS showed a similar pattern with significant surface area reduction in ipsilateral entorhinal cortex, parahippocampal gyrus, temporal pole, superior temporal gyrus, and the anterior parts of the middle and inferior temporal gyri. In addition, smaller localized regions of surface area reduction were detected in ipsilateral insular cortex, ipsilateral superior frontal gyrus, and contralateral precuneus cortex.

Figure 1.


Patterns of cortical surface area contraction in patients with left and right MTLE + HS compared to healthy controls. In each hemisphere, areas of reduced cortical surface area in (A) left (n = 35) and (B) right MTLE + HS patients (n = 35) compared to healthy controls (n = 40) are shown. Maps of reduced surface area are reported at p < 0.05 (survived correction for multiple comparisons). In each panel, the top row represents the left hemisphere and the bottom row represents the right hemisphere. Color bar represents statistical significance: controls > MTLE + HS patients (0.05 < p < 0.0001, red/yellow), patients > controls (0.05 < p < 0.0001, blue/cyan).

Group differences in cortical surface geometric distortion

Figure 2 illustrate changes in metric distortion (indicating altered surface morphology and abnormal cortical folding) in the patient groups compared to the healthy controls. Both left and right MTLE + HS patient groups showed regions of significantly increased metric distortion predominantly within the ipsilateral mesial and anterior temporal lobe subregions. This pattern is remarkably similar to that observed for surface area changes. Increased metric distortion was detected in ipsilateral entorhinal cortex, parahippocampal gyrus, and temporal pole in left and right MTLE + HS patients. Furthermore, patients with left MTLE + HS showed increased metric distortion in contralateral lateral occipital cortex and decreased metric distortion in ipsilateral medial orbitofrontal cortex and contralateral paracentral lobule. Patients with right MTLE + HS also showed increased metric distortion in ipsilateral superior frontal gyrus and contralateral precuneus cortex.

Figure 2.


Patterns of cortical surface geometric distortion in left and right MTLE + HS patients compared to healthy controls. In each hemisphere, areas of increased/decreased metric distortion in (A) left (n = 35) and (B) right MTLE + HS patients (n = 35) compared to healthy controls (n = 40) are shown. Maps of metric distortion are reported at p < 0.05 (survived correction for multiple comparisons). In each panel, the top row represents the left hemisphere and the bottom row represents the right hemisphere. Color bar represents statistical significance: controls > MTLE + HS patients (0.05 < p < 0.0001, red/yellow), patients > controls (0.05 < p < 0.0001, blue/cyan).

Group differences in cortical thickness

Figure 3 illustrates the results of the cortical thickness analysis. Cortical thinning was observed in a pattern different from that observed for surface area and metric distortion. Cortical thinning was detected within regions dispersed across the cortex bilaterally. In left MTLE + HS patients, regions of cortical thinning were identified within the ipsilateral superior temporal gyrus, superior frontal gyrus, precentral gyrus, and lingual gyrus; and bilaterally in regions within the paracentral lobule, superior parietal cortex, and precuneus cortex. In addition, regions of cortical thinning were detected in contralateral inferior frontal gyrus and the isthmus division of the cingulate cortex. In patients with right MTLE + HS, regions of cortical thinning were identified ipsilaterally in the postcentral gyrus and lateral occipital cortex; and contralaterally in middle frontal gyrus and anterior cingulate cortex. Similarly, patients with right MTLE + HS showed thinning regions bilaterally in paracentral lobule, superior parietal cortex, and precuneus cortex; as well as in the transverse temporal gyrus and superior frontal gyrus.

Figure 3.


Patterns of cortical thinning in left and right MTLE + HS patients compared to healthy controls. In each hemisphere, areas of cortical thinning in (A) left (n = 35) and (B) right (n = 35) MTLE + HS patients compared to healthy controls (n = 40) are shown. Maps of cortical thinning are reported at p < 0.05 (survived correction for multiple comparisons). In each panel, the top row represents the left hemisphere and the bottom row represents the right hemisphere. Color bar represents statistical significance: controls > MTLE + HS patients (0.05 < p < 0.0001, red/yellow), patients > controls (0.05 < p < 0.0001, blue/cyan).

The correlation between ipsilateral hippocampal volume and cortical surface area

Given that the hippocampus is recognized as a key structure in seizure generation/propagation in MTLE + HS, we explored the correlation between ipsilateral hippocampal volume and changes in cortical surface area. After controlling for ICV and age, ipsilateral temporal surface area showed a strong positive correlation with ipsilateral hippocampal volume in patients with left and right MTLE + HS. Specifically, hippocampal volume correlated positively with ipsilateral surface area of the entorhinal cortex, parahippocampal gyrus, temporal pole, and middle temporal gyri in both MTLE + HS patient groups. These apparently strong correlations were absent both in healthy controls and on the contralateral side of the same patients (see Fig. 4 and Fig. S1).

Figure 4.


Correlation of cortical surface area with hippocampal volume. In each hemisphere, areas of significant correlations between surface area and hippocampal volume in (A) left (n = 35) and (B) right (n = 35) MTLE + HS patients are shown. Significant correlations are reported at p < 0.05 (survived correction for multiple comparisons). In each panel, the top row represents the left hemisphere and the bottom row represents the right hemisphere. Color bar represents statistical significance: positive correlation (0.05 < p < 0.0001, red/yellow), negative correlation (0.05 < p < 0.0001, blue/cyan).

Correlation of clinical features with cortical surface area

To assess the influence of disease-related factors on cortical surface area, we explored the relationship between cortical surface area and disease duration, age at seizure onset, and estimates of lifetime number of partial and generalized seizures. After controlling for ICV and age, disease duration correlated negatively with ipsilateral anterior and mesial temporal subregional surface area. Results are shown Fig. 5. This correlation appeared stronger in patients with left MTLE + HS. In a very similar pattern, age at seizure onset correlated positively with ipsilateral temporal subregions. This is likely explained by the significant correlation between disease duration and age at seizure onset (r = −0.63). No significant correlations were observed between cortical surface area and estimates of lifetime numbers of partial or generalized seizures.

Figure 5.


Correlation of cortical surface area with duration of epilepsy. In each hemisphere, areas of significant correlations between surface area and duration of epilepsy in (A) left (n = 35) and (B) right (n = 35) MTLE + HS patients are shown. Significant correlations are reported at p < 0.05 (survived correction for multiple comparisons). In each panel, the top row represents the left hemisphere and the bottom row represents the right hemisphere. Color bar represents statistical significance: positive correlation (0.05 < p < 0.0001, red/yellow), negative correlation (0.05 < p < 0.0001, blue/cyan).

Given the hypothesis that IPIs are the driver of progressive atrophy in MTLE + HS (Meyer et al., 1954; Mathern et al., 2002), we tested the influence of IPIs on surface area by comparing cortical surface area in patients with a positive history to those with a negative history of IPIs. Our results showed no clear IPI-specific effect on cortical surface area (see Fig. S2).

Discussion

To our knowledge, this study is the first to report characteristic changes in cortical surface area in MTLE + HS. The findings illustrate asymmetric patterns of cortical surface area changes with reductions predominating in ipsilateral mesial and anterior temporal lobe subregions. Patterns of cortical surface area reductions were distinct from neocortical thinning, which was evident in localized regions dispersed throughout the cortex bilaterally. These distinct patterns of cortical surface area contractions and neocortical thinning likely reflect their independence as biomarkers of different neuropathologic mechanisms. The significant overlap between surface geometric distortion and surface area findings suggests that areal contraction is a significant component of the morphologic alterations in MTLE. Furthermore, evidence of a strong correlation between hippocampal volume and temporal lobe subregional surface area was detected and was restricted to the ipsilateral side of seizure activity in patients. Ipsilateral temporal lobe surface area also correlated negatively with disease duration. These correlations suggest a progressive reduction in surface area and implicate the sclerotic hippocampus.

Neuropathologic and MRI-based studies of patients with MTLE + HS have consistently reported extrahippocampal cortical abnormalities in several limbic and neocortical regions, primarily ipsilateral to the side of seizure onset. Evidence of “extrahippocampal” damage has also been demonstrated in animal models of MTLE involving seizure-induced hippocampal injury (Schauwecker, 2003). In patients with refractory MTLE + HS, structural abnormalities have been described in ipsilateral temporal lobe subregions known to be anatomically connected to the hippocampus. For example, volume reduction has been revealed in the entorhinal cortex (Bernasconi et al., 1999), parahippocampal gyrus (Bernasconi et al., 2003), and temporal pole (Moran et al., 2001). Gray matter volume (GMV), however, reflects a combination of cortical thickness and surface area measurements, which likely echo distinct neurobiologic and genetic processes (Panizzon et al., 2009). Cortical GMV alterations cannot be attributed to a single pathologic process. Distinguishing cortical surface area and thickness changes may reveal distinct cortical abnormalities and thus help to further elucidate cortical pathogenesis in MTLE + HS. Cortical thinning has been described previously in lateral temporal regions (Lin et al., 2007; McDonald et al., 2008b); however, it has not fully explained volume deficits in mesial temporal regions. Our findings of reduced cortical surface area in mesial and anterior subregions of the ipsilateral temporal lobe indicate that contractions in surface area, rather than cortical thinning, are the main driver of volume deficits in these cortical regions.

Although the underlying biologic correlates of local cortical surface area are poorly understood, the measurement reflects the overall degree of cortical folding (Im et al., 2008). During normal neurodevelopment and brain maturation, differential growth of cortical layers (Armstrong et al., 1995) and the establishment of interregional axonal connections (Van Essen, 1997) change the smooth fetal brain to a more complex and folded adult human brain. Cortical folding is believed to reflect the state of interregional axonal connections and cortical connectivity (Van Essen, 1997). It is possible that the asymmetric surface area and morphology changes identified in ipsilateral temporal lobe subregions reflect changes in underlying temporal lobe connectivity and white matter tracts. This is supported by the consistent finding of ipsilateral temporal white matter abnormalities in MTLE + HS (Ahmadi et al., 2009). Evidence of a relationship between cortical surface morphologic features (e.g., cortical folding) and aberrant patterns of cortical connectivity has been suggested by several studies of different neurodevelopmental disorders (White et al., 2003; Hardan et al., 2004).

The epileptogenic zones in MTLE + HS are believed to extend beyond the sclerotic hippocampus. Interactions among the hippocampus, entorhinal cortex, and temporal pole play a significant role in generation and propagation of some seizure types in MTLE (Spencer & Spencer, 1994; Bartolomei et al., 2005; Chabardès et al., 2005). These interactions are likely related to the known anatomic pathways connecting these regions and may explain some failures of selective mesial temporal lobe resections (Berkovic et al., 1995; Chabardès et al., 2005). The strong correlations we observed between temporal lobe subregional surface area and ipsilateral hippocampal volume support a role for compromised temporal lobe networks and may indicate a common pathologic process responsible for HS and a network of regions distributed within the temporal lobe (Moran et al., 2001). Alternatively, they may suggest a combination of (1) an injury extending from the hippocampus (e.g., seizure-induced injury) and (2) deafferentation of hippocampal pathways. Further work is required to determine the significance of these correlations.

Alterations to cerebral cortex surface features are thought to result secondary to interruptions of normal cortical development by genetic or early environmental factors (Armstrong et al., 1995). The genetic determinants of cortical surface area are thought to be linked to tangential migration of neurons during early cortical development (Rakic, 1988). Therefore, our current surface area observations may be linked to area-specific genetic markers that are highly relevant to the pathophysiology of MTLE + HS. Alternatively, it is possible that surface abnormalities in ipsilateral temporal lobe are acquired and related to early neurodevelopmental factors. The IPIs we tested did not appear to influence surface area changes, although environmental factors other than those we investigated may be contributing. Although our sample included a small number of drug-responsive patients (n = 9), our results are more supportive of progressive changes in surface area relating to disease burden as suggested by the observed negative correlation between temporal lobe subregional surface area and disease duration.

In this study, the patterns of cortical surface area and surface geometric distortion alterations were distinct from that of cortical thinning. Previous work has described bilateral cortical thinning in both patients with mild and refractory MTLE, affecting all cerebral lobes, preferentially the sensorimotor cortices (Lin et al., 2007; McDonald et al., 2008a; Bernhardt et al., 2009; Labate et al., 2011). Our cortical thickness findings are in line with these previous reports. The underlying pathogenesis of such cortical thinning is not yet fully understood, but most of these changes have been attributed to neuronal injury to cortical regions involved in the epileptogenic network (Bernhardt et al., 2009; Labate et al., 2011).

Despite the interesting findings we have presented, we recognize that there are limitations to this study. In terms of MR image acquisition, it is usually recommended that two T1-weighted scans are acquired and averaged for best results from FreeSurfer cortical reconstruction. This was not possible due to the clinical setting of our study. Our sample size was modest. Additional cases or indeed a larger control group would have provided more statistical power. Finally, to control for heterogeneity in the patient sample, we included only patients with MTLE + HS. Future work is needed to explore surface area changes in MTLE without hippocampal sclerosis.

Acknowledgments

The authors thank all patients and participants who took part in this study. This work was funded by Science Foundation Ireland Research Frontiers Programme award 08/RFP/GEN1538.

Disclosure

The authors have no conflicts 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.

Ancillary