Increased temporolimbic cortical folding complexity in temporal lobe epilepsy

Abstract

Objective:

Converging evidence suggests that abnormalities of brain development may play a role in the pathogenesis of temporal lobe epilepsy (TLE). As sulco-gyral patterns are thought to be a footprint of cortical development, we set out to quantitatively map folding complexity across the neocortex in TLE. Additionally, we tested whether there was a relationship between cortical complexity and features of hippocampal maldevelopment, commonly referred to as malrotation.

Methods:

To quantify folding complexity, we obtained whole-brain surface-based measures of absolute mean cortical curvature from MRI scans acquired in 43 drug-resistant patients with TLE with unilateral hippocampal atrophy, and 40 age- and sex-matched healthy controls. In patients, we correlated changes in cortical curvature with 3-dimensional measures of hippocampal positioning.

Results:

We found increased folding complexity in the temporolimbic cortices encompassing parahippocampal, temporopolar, insular, and fronto-opercular regions. Increased complexity was observed ipsilateral to the seizure focus in patients with left TLE (LTLE), whereas these changes were bilateral in patients with right TLE (RTLE). In both TLE groups, increased temporolimbic complexity was associated with increased hippocampal malrotation. We found tendencies for increased complexity in bilateral posterior temporal cortices in LTLE and contralateral parahippocampal cortices in RTLE to be predictive of unfavorable seizure outcome after surgery.

Conclusion:

The anatomic distribution of increased cortical complexity overlapping with limbic seizure networks in TLE and its association with hippocampal maldevelopment further imply that neurodevelopmental factors may play a role in the epileptogenic process of TLE.

Growing evidence suggests abnormalities of brain development are etiologic factors contributing to the pathogenesis of temporal lobe epilepsy (TLE). Although hippocampal sclerosis is the histopathologic hallmark of this condition, detailed studies have shown concurrent abnormal dispersion of granule cells, remnant fetal Cajal-Retzius cells, and changes in architectural organization of CA1 and dentate gyrus indicative of underlying hippocampal developmental abnormalities.¹ Additional data revealed increased temporal lobe white matter neuronal clustering² and abnormal cortical myelinated fibers resembling those found in malformations of cortical development.^3,4 While current MRI techniques cannot readily identify these small-scale features, macroscopic markers of hippocampal maldevelopment (referred to as malrotation), identified as atypical shape and positioning, are found in 40% of patients with TLE.^5,6

The anatomic location and extent of potential cortical developmental anomalies in TLE remain largely unknown. Sulco-gyral patterns are thought to be a footprint of cortical development. Previous studies using automated measures of gyrification in TLE reported divergent findings, including increased gyral curvature,⁷ decreased fractal dimension,⁸ and no significant changes.⁹ Notably, these studies used different indexes of folding complexity and provided only a single measure at the lobar or whole-brain level, lacking sensitivity to focal changes.

Our objectives were to map absolute cortical curvature as a measure of folding complexity at every point of the neocortex in drug-resistant patients with TLE, test for associations between measures of cortical complexity and hippocampal maldevelopment, and investigate the relationship between these structural measures and clinical parameters, including surgical outcome.

METHODS

Subjects.

Based on manual hippocampal volumetry,¹⁰ we selected from our database 43 right-handed patients (19 men, mean age = 35 ± 11 years, range = 18–63 years) with drug-resistant TLE and unilateral hippocampal atrophy, determined as hippocampal volumes or an interhemispheric hippocampal asymmetry beyond 2 SD of the corresponding mean of healthy controls.

TLE diagnosis and lateralization of the seizure focus into left TLE (LTLE, n = 22) and right TLE (RTLE, n = 21) were determined by comprehensive evaluation including seizure history and semiology, video-EEG telemetry, and neuroimaging in all patients. None of the patients had a mass lesion (tumor, vascular malformation, or malformations of cortical development) or traumatic brain injury.

Thirty-seven patients (19 LTLE, 18 RTLE) underwent a selective amygdalohippocampectomy. We determined surgical outcome according to the Engel classification scheme¹¹ at a mean follow-up time of 4 ± 3 years. Twenty-five (69%) patients had an outcome Class I (20 Class Ia), 5 (14%) Class II, 3 (8%) Class III, and 3 (8%) Class IV. One patient was lost to follow-up. Following qualitative histopathologic analysis, hippocampal sclerosis was detected in all 22 patients in whom hippocampal specimen was available. In the remaining 15, hippocampal specimens were either incomplete (n = 7) or unsuitable for histopathology due to subpial aspiration (n = 8).

The control group consisted of 40 age- and sex-matched right-handed healthy volunteers (17 men, mean age = 33 ± 11 years, range = 20–66 years). Demographic and clinical data for all subjects are presented in the table.

Table.

Demographic and clinical data

Abbreviations: Engel I = seizure-free, i.e., Class I postsurgical outcome in Engel classification; FC = febrile convulsions; LTLE = left temporal lobe epilepsy; RTLE = right temporal lobe epilepsy.

Standard protocol approvals, registrations, and patient consents.

The Ethics Committee of the Montreal Neurological Institute and Hospital approved the study and written informed consent was obtained from all participants.

MRI data acquisition.

T1-weighted whole-brain structural images were acquired on a 1.5 T Gyroscan magnetic resonance scanner (Philips Medical Systems, Eindhoven, the Netherlands) using a 3-dimensional T1-fast field echo sequence (repetition time = 18 msec; echo time = 10 msec; number of excitations = 1; flip angle = 30°; matrix size = 256 × 256; field of view = 256 × 256 mm²; slice thickness = 1 mm), providing an isotropic voxel size of 1 × 1 × 1 mm.

MRI processing and cortical surface extraction.

Preprocessing involved automated correction for intensity nonuniformity and intensity standardization, linear registration into standardized stereotaxic space based on the Talairach atlas, and automatic tissue classification into white matter, gray matter, and CSF. We applied the Constrained Laplacian Anatomic Segmentation using Proximity algorithm¹² to generate a model of the inner (white matter) and outer (gray matter) surfaces with 40,962 surface points or vertices for each hemisphere. Extracted surfaces were nonlinearly aligned to a surface template using a 2D registration procedure that improves interindividual anatomic correspondence.¹³ The accuracy of gray and white matter surface extractions was verified in all subjects prior to further analysis.

Cortical curvature measurement.

For cortical curvature analyses, we generated a midsurface model representing the mid-distance between the inner (gray matter–white matter) and outer (gray matter–CSF) surfaces. Curvature measurements along the midsurface are less biased to sulcal or gyral regions than sampling either the inner or outer surfaces.¹⁴ We subsequently applied a barycentric smoothing with 3 iterations to reduce high-frequency noise in the vertex positions.¹⁵ Absolute mean curvature was calculated at each vertex as a measure of changes in the frequency and depth of sulcal and gyral folds, thus expressing the local amount of gyrification or complexity.¹⁶ Mean curvature data were blurred using smoothed kernels of 10, 20, and 30 mm full width at half maximum.

Quantitative analysis of hippocampal maldevelopment.

We chose to study vertical orientation and medial positioning, the most commonly described macroscopic features of hippocampal malrotation.^5,6 As these features characterize the fetal hippocampus until approximately 18 weeks of gestation,¹⁷ their persistence into adult life is considered to represent hippocampal maldevelopment. Indeed, normal development entails a displacement of the hippocampus toward the lateral portion of the choroid fissure (into the temporal lobe medial wall), and its shifting to a nearly horizontal orientation. To quantify these characteristics, 3-dimensional models of the left and right hippocampi in each subject were created from manual labels¹⁸ to determine 1) sagittal translation, measuring the position of the hippocampus relative to the midline; 2) axial rotation, reflecting a medial-lateral deflection of the hippocampus relative to its geometric center; and 3) longitudinal rotation, indicating a relative vertical deviation of the entire hippocampus from its normally horizontal orientation. Sagittal translation was calculated as millimeter distance between the geometric center of the hippocampus and the midsagittal plane of the brain. Axial rotation and longitudinal rotation were measured in angles of deviation from the orientation of a reference average hippocampus, which was manually segmented on the ICBM-152 template. These measurements were normalized through a z transformation relative to the corresponding distribution of controls.

Statistical analysis.

Surface-based analyses were performed using the SurfStat (http://galton.uchicago.edu/∼worsley/surfstat) toolbox for Matlab (R2007a, The Mathworks, Natick, MA). Results are reported based on curvature data after 30 mm smoothing, unless otherwise specified.

We compared cortical curvature measurements between groups using vertex-wise t tests. Resulting t statistic maps were corrected for multiple comparisons using random field theory at p < 0.05 on a cluster level. In significant clusters, curvature measurements were normalized through a z transformation relative to the corresponding distribution of controls to determine the proportion of individuals with abnormal folding.

As curvature may be affected by changes in cortical thickness, we measured thickness as previously described.^19,20 We then repeated the curvature analysis, adjusting for potential thickness confounds at every vertex.

In patients with TLE, linear models were employed to determine the relationship between z-normalized cortical curvature and hippocampal malrotation features, as well as clinical measures including age, age at seizure onset, duration of epilepsy, and surgical outcome.

RESULTS

Cortical curvature analysis.

Results are shown in figure 1. Compared to healthy controls, patients with LTLE and patients with RTLE showed clusters of increased cortical curvature in the ipsilateral mesial temporal lobe involving the temporopolar and parahippocampal regions (p < 0.01). In patients with RTLE, the increased curvature extended to the posterior cingulate cortices (p < 0.01). In these regions, high proportions of patients with TLE, but not controls, had a curvature z score greater than 1.5 (LTLE: 64% [14 of 22]; RTLE: 48% [10 of 21]; controls: 2.5% [1 of 40]). Patients with RTLE also showed increased complexity in the contralateral left temporopolar region (p < 0.01). In addition, both patients groups displayed left hemispheric clusters of increased complexity involving the fronto-opercular and insular cortices (p < 0.01). In these regions, 64% (14 of 22) of patients with LTLE and 48% (10 of 21) of patients with RTLE had a curvature z score greater than 1.5, while this was seen in only 10% (4 of 40) of controls. We did not find any regions of decreased curvature in patients compared with controls.

Figure 1. Vertex-wise group analysis of cortical folding complexity.

Regions of significantly increased absolute mean curvature in (A) left temporal lobe epilepsy (LTLE) and (B) right temporal lobe epilepsy (RTLE) relative to healthy controls are displayed at 30-mm smoothing kernels. Significant clusters (t > 2.0, extent > 3.0 resels) thresholded using random field theory (rft) are outlined in black. Results are also displayed at an uncorrected threshold of p < 0.025.

These results were consistent across 10- and 20-mm full width at half maximum smoothing kernels and remained unchanged when correcting for cortical thickness at every vertex.

Relationship between curvature and hippocampal malrotation.

We first tested for group differences in hippocampal positioning. Comparing patients to controls, we observed ipsilateral increased longitudinal rotation (t = 4.59, p < 0.001) and sagittal translation (t = 2.07, p < 0.05) in LTLE. These features were increased bilaterally in patients with RTLE (longitudinal rotation: right: t = 2.47, p < 0.02; left: t = 2.26, p < 0.03; sagittal translation: right: t = 4.03, p < 0.001; left: t = 2.69, p < 0.001).

We subsequently assessed the relationship between hippocampal longitudinal rotation and sagittal translation and cortical folding. In LTLE, increased folding in the left mesial temporal (i.e., temporopolar and parahippocampal) and insulo-opercular cortices was related to an increase in left hippocampal longitudinal rotation (r = 0.45, p = 0.03, figure 2). In RTLE, increased folding of the right mesial temporal cortices (i.e., temporopolar and parahippocampal) was associated with increased right hippocampal sagittal translation (r = 0.55, p = 0.008).

Figure 2. Correlation analysis between cortical folding complexity and hippocampal longitudinal rotation in temporal lobe epilepsy (TLE).

The scatterplot displays a positive association between left hemispheric z scores of hippocampal longitudinal rotation and mesiotemporal and insulo-opercular mean cortical curvature in patients with left TLE. Two examples drawn from the correlation analysis are shown in the lower panels.

Relationship between curvature and clinical measures.

In relation to postsurgical outcome, patients with RTLE with residual seizures (Engel II–IV, n = 7) had a tendency (p < 0.025, uncorrected) for increased folding in the left anterior parahippocampal cortex compared to patients who became seizure-free (Engel I, n = 11). Patients with LTLE with residual seizures (Engel II–IV, n = 4) showed trends (p < 0.025, uncorrected) for bilateral increased curvature in posterior temporal and ipsilateral insular cortices compared to patients who became seizure-free (Engel I, n = 14).

No relationship was found between curvature and age at scan, age at onset of seizures, or duration of seizures.

DISCUSSION

We identified areas of increased folding complexity in the temporolimbic cortices encompassing parahippocampal, temporopolar, insular, and fronto-opercular regions. In patients with LTLE, increased complexity was observed mainly ipsilateral to the seizure focus, whereas changes were bilateral in patients with RTLE. Our findings were robust to different smoothing kernels and independent of concurrent changes in cortical thickness. In both TLE groups, areas of increased complexity were related to hippocampal malrotation. We found that tendencies for increased complexity in bilateral posterior temporal cortices in LTLE and contralateral parahippocampal cortices in RTLE were predictive of unfavorable seizure outcome after surgery. However, the significance of this finding remains to be determined in a larger series of patients with balanced numbers for each outcome group.

Using a region of interest–based index, we previously showed ipsilateral increased curvature of the superior temporal gyrus in TLE.²¹ This initial observation has since been extended in a few automated whole-brain studies of cortical complexity, with inconsistent results.^7–9 While our present findings appear at odds with the decreased multilobar fractal dimension previously reported,⁸ the nonspecific nature of the fractal dimension measurement limits a direct comparison with our method. Another publication reported a nonlocalized trend toward increased folding as assessed by isoperimetric ratio.⁹ While this study would support our findings, it might have had limited power associated with a small sample size. Moreover, only half of the patients had mesiotemporal sclerosis. The direction of our findings favoring increased sulco-gyral folding complexity is in agreement with data from another study,⁷ although the exact anatomic localization was not provided.

During the initial phase of brain development, the cerebral surfaces are smooth. Most sulci and gyri develop during the third trimester, and the primary and secondary fissures are visible at birth.²² Typically, the degree of gyrification stabilizes during early childhood.²³ The hippocampus undergoes a similar process during which it evolves from an unfolded smooth aspect to a more complex and folded structure.¹⁷ The processes leading to cortical and hippocampal folding are possibly similar, as suggested by the association of lissencephaly with hippocampal malrotation. These processes likely involve mechanical forces resulting from different tension of growth between early cortical strata,²⁴ tension along axons between interactive cortical areas,²⁵ and changes in subcortical connectivity.²⁶ We propose that early fetal pathologic mechanisms affecting hippocampal development may be among the factors triggering anomalies in areas receiving dense projections from the hippocampus, such as the parahippocampal, temporopolar, and interconnected insular and orbitofrontal cortices,^27,28 consequently altering their gyrification process. This hypothesis is also supported by histologic samples in TLE showing white matter abnormalities unrelated to gliosis,²⁹ as well as abnormal axon bundles potentially representative of aberrant cortico-cortical projection fibers akin to those observed in malformations of cortical development.³ Finally, the frequent co-occurrence of hippocampal sclerosis in patients with focal cortical dysplasia³⁰ (known as dual pathology) and atypical basal temporal lobe sulcal arrangement on MRI³¹ in patients with TLE lend support to a neurodevelopmental problem underlying abnormal cortical folding in TLE. Future studies aiming at assessing the relationship between cortical folding and large-scale alterations in white matter pathway organization^32,33 would further advance our understanding of developmental aspects of TLE. Although the presence of disease-related increased cortical folding does not conclusively prove a neurodevelopmental abnormality, the latter is in keeping with the relationship we observed between hippocampal malrotation features and cortical folding in our patients. The fact that we have previously observed equal proportions of malrotation features in patients with TLE with hippocampal atrophy and those with normal hippocampal volume⁶ minimizes the likelihood for potential interference of atrophic changes with our measurements.

While we cannot rule out the contribution of early injuries to our present findings, previous work has suggested that febrile seizures result from—rather than predispose to—hippocampal malformations,³⁴ supporting the notion that TLE may result from a preexisting neurodevelopmental abnormality.³⁵

Our observation of a bilateral albeit asymmetric increase in cortical folding complexity and hippocampal malrotation features^6,36 in patients with RTLE supports the intriguing possibility of hemisphere-specific vulnerability to injury possibly related to asymmetric brain development (see ³⁷ for review). In this scenario, abnormalities would occur during a time window of left hemisphere susceptibility when right hemisphere homologues, due to a more advanced stage of development, are relatively spared.

Noticeable alterations of folding patterns in epilepsy are usually found in patients with epileptogenic malformations of cortical development, ranging from reduction in gyration, such as that seen in the lissencephaly complex, to increased folding frequency in polymicrogyrias.³⁸ Since in the absence of any obvious malformation unusual folding patterns are thought to be a marker of subtle cortical dysgenesis, our findings suggest that neurodevelopmental factors may play a role in the emergence of epileptogenesis in TLE. Indeed, the anatomic distribution of folding anomalies in our patients, involving the temporopolar, parahippocampal, and insulo-opercular cortices, overlaps notably with limbic seizure networks in TLE.^39,40

DISCLOSURE

Dr. Voets was previously employed by GlaxoSmithKline, in which she currently holds stock, and has received a Centers of Excellence for Commercialization and Research (CECR) Fellowship. B.C. Bernhardt received a JTC fellowship from the Montreal Neurological Institute and a studentship from the Savoy Foundation for Epilepsy. H. Kim, Dr. Yoon, and Dr. Bernasconi report no disclosures.