SEARCH

SEARCH BY CITATION

Keywords:

  • Aging;
  • Temporal lobe epilepsy;
  • MRI;
  • Volumetrics

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

Purpose:  To characterize differences in brain structure and their patterns of age-related change in individuals with chronic childhood/adolescent onset temporal lobe epilepsy compared with healthy controls.

Methods:  Subjects included participants with chronic temporal lobe epilepsy (n = 55) of mean childhood/adolescent onset and healthy controls (n = 53), age 14–60 years. Brain magnetic resonance imaging (MRI) studies (1.5 T) were processed using FreeSurfer to obtain measures of lobar thickness, area, and volume as well as volumes of diverse subcortical structures and cerebellum. Group differences were explored followed by cross-sectional lifespan modeling as a function of age.

Key Findings:  Anatomic abnormalities were extensive in participants with chronic temporal lobe epilepsy including distributed subcortical structures (hippocampus, thalamus, caudate, and pallidum), cerebellar gray and white matter, total cerebral gray and white matter; and measures of cortical gray matter thickness, area, or volume in temporal (medial, lateral) and extratemporal lobes (frontal, parietal). Increasing chronologic age was associated with progressive changes in diverse cortical, subcortical, and cerebellar regions for both participants with epilepsy and controls. Age-accelerated changes in epilepsy participants were seen in selected areas (third and lateral ventricles), with largely comparable patterns of age-related change across other regions of interest.

Significance:  Extensive cortical, subcortical, and cerebellar abnormalities are present in participants with mean chronic childhood/adolescent onset temporal lobe epilepsy implicating a significant neurodevelopmental impact on brain structure. With increasing chronologic age, the brain changes occurring in epilepsy appear to proceed in a largely age-appropriate fashion compared to healthy controls, the primary exception being age-accelerated ventricular expansion (lateral and third ventricles). These cumulative structural abnormalities appear to represent a significant anatomic burden for persons with epilepsy, the consequences of which remain to be determined as they progress into elder years.

The issue of how chronic temporal lobe epilepsy (TLE) affects brain structure and cognition across the lifespan remains a topic of interest and concern. Tightly controlled prospective investigations of cognition in epilepsy are modest in number and most have a test–retest interval of well below 10 years (Dodrill, 2004; Seidenberg et al., 2007). An emerging view derived from the combination of cross-sectional and prospective studies is that childhood/adolescent onset epilepsy exerts a significant adverse neurodevelopmental effect on cognition and that, over time, persons with persisting pharmacoresistent chronic epilepsy carry their deficits into and through adulthood. A subset of patients exhibit an especially difficult prospective cognitive course—those problems less linked to recorded seizure frequency than baseline sociodemographic characteristics (e.g., age, education, intelligence quotient [IQ]), neuroimaging abnormalities, or the underlying etiology of the epilepsy (Dikmen et al., 1975; Dodrill & Matthews, 1992; Helmsteadter et al., 2003; Piazzini et al., 2006; Hermann et al., 2008; Kaaden & Helmstaedter, 2009; Helmstaedter & Elger, 2009; Baxendale et al., 2010).

With respect to prospective changes in brain structure, a largely comparable state of affairs exists with a limited number of controlled longitudinal investigations, again typically with modest test–retest intervals (e.g., Liu et al., 2003; Liu et al. 2005; Bernhardt et al., 2009; Coan et al., 2009; Bernhardt et al., 2010). Several prospective studies have focused on discrete but important regions of interest in specific epilepsy syndromes, such as the hippocampus in TLE (e.g., Van Paesschen et al., 1998; Briellmann et al., 2002; Fuerst et al., 2003). A number of cross-sectional investigations of prevalent cases have characterized distributed abnormalities in cortical volume or thickness, volumes of subcortical structures and cerebellum (Sisodiya et al., 1997; Bernasconi et al., 2004; Mueller et al., 2006; Lin et al., 2007; Keller & Roberts, 2008; McDonald et al., 2008a,b; Hermann et al., 2009; Bonilha et al., 2010a,b), as well as white matter volume and other measures of connectivity in patients with mesial TLE (Arfanakis et al., 2002; Hermann et al., 2003a,b; Diehl et al., 2008; McDonald et al., 2008a,b,c; Concha et al., 2009; Meng et al., 2010; Riley et al., 2010). How these abnormalities change with age over the broader lifespan is unknown.

In the literature on normal aging, a number of quantitative neuroimaging investigations have examined healthy controls across a broad age range to provide preliminary cross-sectional insight into lifespan trajectories of brain change (e.g., Pfefferbaum et al., 1994; Blatter et al., 1995; Toga et al., 2006; Sowell et al., 2007; Raz et al., 2010; Walhovd et al., 2011). This modeling approach has been used to examine lifespan cognition in epilepsy (Helmstaedter & Elger, 2009; Baxendale et al., 2010), but has not yet attempted to characterize lifespan brain structure in epilepsy. Herein we report the results of a cross-sectional lifespan analysis of brain structure among adult patients with chronic TLE of mean childhood/adolescent onset compared to healthy controls in order to address three questions: (1) what is the broad landscape of anatomic abnormality in middle-aged persons with mean childhood/adolescent onset TLE, (2) what are the patterns of change in cortical and subcortical structure as a function of chronologic age in the epilepsy and control groups, and (3) are age-accelerated brain changes evident in the epilepsy group?

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

Subjects

Research participants were patients with TLE (n = 55) and healthy controls (n = 53).

Initial selection criteria for epilepsy patients included: (1) chronologic age from 14–60 years, (2) complex partial seizures of definite or probable temporal lobe origin (see details below), (3) absence of magnetic resonance imaging (MRI) abnormalities other than atrophy on clinical reading, and (4) no other neurologic disorder. Epileptologists reviewed patients’ medical records including seizure semiology and previous electroencephalography (EEG) and neuroimaging reports. Each patient was classified as having seizures of definite, probable, or possible temporal lobe origin. Definite TLE was defined by continuous video-EEG monitoring of spontaneous seizures demonstrating temporal lobe seizure onset; probable TLE was determined by review of clinical semiology with features reported to reliably identify complex partial seizures of temporal lobe origin versus onset in other regions (e.g., frontal lobe) in conjunction with interictal EEG studies, neuroimaging findings, and developmental and clinical history. Only patients meeting criteria for definite and probable TLE proceeded to recruitment for study participation; patients with possible TLE were excluded. Because ictal EEG monitoring is the gold standard for localization and lateralization of seizure onset, secondary analyses were computed using only these patients. The results to be described were confirmed in all cases when analyses were limited to patients with ictally confirmed left and right TLE, and these findings are presented in the Results section.

Selection criteria for healthy controls included (1) chronologic age from 14–60; (2) either a friend or family member of the patient in order to maximize sociodemographic similarity; (3) no current substance abuse, medical, or acute psychiatric condition that could affect cognitive functioning; and (4) no history of loss of consciousness >5 min or developmental learning disorder. All subjects (epilepsy and controls) were excluded if Full Scale IQ was <70. Subject characteristics are provided in Table 1.

Table 1.   Demographic characteristics
 Controls (n = 53)Epilepsy (n = 55)
Age31.4 (11.5)35.6 (11.7)
Gender17M, 35F14M, 41F
Onset (mean years) 14.9
Duration (mean years) 21.0
Seizure frequency (%)  
 Controlled 16
 Daily 8
 Weekly 29
 Monthly 39
 Yearly 8
Number of medications (%)  
 One 47
 Two 40
 Three 13

MRI procedures

MRI acquisition

Images were obtained on a 1.5 T GE Signa MR scanner (GE Medical Systems, Milwaukee, WI, U.S.A.). The sequence acquired for each participant was a T1-weighted, three-dimensional spoiled gradient (SPGR) acquired with the following parameters: echo time (TE) = 5, repetition time (TR) = 24, flip angle = degrees, one excitation, slice thickness = 1.5 mm, slices = 124, plane = coronal, field of view (FOV) = 200 mm, matrix = 256 × 256.

All MR images are inspected before image processing. Image quality was rated on a five-point scale as described at (http://www.psychiatry.uiowa.edu/mhcrc/IPLpages/qa_main.htm). We required a minimum of quality 3 or better for the scan to be included in this analysis.

MRI processing

Images were transferred to a Mac OSX (Apple Inc., Cupertino, CA, U.S.A.) computer for processing with the FreeSurfer image analysis suite which is documented and freely available for download online (http://surfer.nmr.mgh.harvard.edu/). FreeSurfer is a set of software tools for the study of cortical and subcortical anatomy. The T1 volumetric MRI scan was used for cortical reconstruction and volumetric segmentation. The technical details of these procedures are described in prior publications (Dale & Sereno, 1993; Dale et al., 1999; Fischl & Dale, 2000; Fischl et al., 2001, 2004b; Segonne et al., 2004; Han et al., 2006; Jovicich et al., 2006). Briefly, this processing includes removal of nonbrain tissue using a hybrid watershed/surface deformation procedure (Segonne et al., 2004), automated Talairach transformation, segmentation of the subcortical white matter and deep gray matter volumetric structures (including hippocampus, amygdala, caudate, putamen, ventricles) (Fischl et al., 2002, 2004a), intensity normalization (Sled et al., 1998), tessellation of the gray matter–white matter boundary, automated topology correction (Fischl et al., 2001; Segonne et al., 2007), and surface deformation following intensity gradients to optimally place the gray–white and gray–cerebrospinal fluid (CSF) borders at the location where the greatest shift in intensity defines the transition to the other tissue class (Dale & Sereno, 1993; Dale et al., 1999; Fischl & Dale, 2000).

After creation of the white matter and pial surfaces, the cortical thickness is calculated as the closest distance from the gray–white boundary to the gray–CSF boundary at each vertex on the tessellated surface (Fischl & Dale, 2000). Surface curvature is also calculated delineating gyral and sulcal features. Comparison of surface measures across subjects is possible after surface inflation (Fischl et al., 1999a), and registration to a spherical atlas, which uses individual cortical folding patterns to match cortical geometry across subjects (Fischl et al., 1999b). The spherical atlas naturally forms a coordinate system in which point-to-point correspondence between subjects can be achieved. Mean thickness differences between groups can then be displayed on the pial surface of the standard atlas. An automatic parcellation of the cerebral cortex into units based on gyral and sulcal structure (Fischl et al., 2004b; Desikan et al., 2006) results in 34 neuroanatomic regions per hemisphere (4 medial temporal, 5 lateral temporal, 11 frontal, 5 parietal, 4 occipital, and 4 cingulate). Average thickness measures are produced for each of the parcellation regions of interest. These 34 regions were combined into six lobar regions per hemisphere for initial analyses. The average thickness for each lobe was calculated as the area weighted average of the lobes constituent regions (mean regional thickness × area for that region +…)/total area of lobe.

The reported data were produced using the FreeSurfer 4.0.4 default processing stream (recon-all -all). Data were visually inspected, and manual interventions were performed when the automated steps reported errors. These included manual alignment to the Talairach template in cases where automated registration was poor, and adjustments to the watershed threshold to restore areas of the brain that were erroneously removed during skull stripping. Occasional editing of the gray matter and white matter segmentation image was completed when defects were evident.

Analyses

The MRI analysis and presentation plan closely followed the report by Ostby et al. (2009) with one significant difference, Ostby et al. focused on the relationship between age and brain development, whereas our focus was largely on differences in later age–related trajectories between healthy controls and participants with epilepsy. Dependent measures included quantitative indices of volume for cerebral cortex and white matter, cerebellar gray and white matter, multiple subcortical structures (hippocampus, amygdala, caudate, pallidum, putamen, thalamus), brainstem and ventricles (lateral, inferior lateral, third, and fourth). In addition, lobar analyses focused on the volume, area, and thickness of the frontal, parietal, occipital, medial and lateral, and cingulate areas.

Three sets of volumetric variables were created including raw volumes, residual volumes after correcting for height and gender using linear regression analyses, and volume as a percentage of total brain volume (TBV) (not shown but available from authors). Surface area and thickness were analyzed using raw measures only. Each measure was analyzed as a standardized z-score to facilitate comparisons across brain regions.

Raw volumes provide a direct measure of the actual differences in volume between groups as well as over life course. There is significant variation across subjects in brain size that would occur even if epilepsy were not a factor. The height- and gender-adjusted volumetric measures account for this variation in brain size and provide greater precision for evaluating the effects of epilepsy on brain structure and age-related changes. Variation across subjects in brain size was accounted for by controlling for height and gender, rather than the more standard method of controlling for intracranial volume (ICV) or TBV, because preliminary analyses revealed significantly smaller ICVs and TBVs among subjects with epilepsy. We view this difference not as a potential confounder of group differences, but as a result of the same process that generates the overall developmental differences for subjects with epilepsy. For this reason, adjusting for either of these measures would lead to an underestimation of the effect of epilepsy on brain structure. In contrast, although the healthy controls were on average 1.1 inches taller than the epilepsy subjects and were more likely to be male (32% vs. 25%), neither of these differences reached statistical significance (p = 0.12 and p = 0.45, respectively) and were therefore not confounded with group membership. They were, however, significantly related to ICV and were therefore used as covariates. Finally, several regions showed evidence of nonlinear change over time. In these cases the natural log of the measure was used in the analysis. Findings regarding raw and gender-/height-adjusted volumes are presented in the text, whereas findings regarding percentage of TBV are available from the authors.

The analytic plan involved multiple comparisons. To preserve an overall p-value of 0.05, an adjustment of the p-value threshold was made. The average correlation between the 16 volumetric brain regions, ventricles, and ICV measures was 0.32, whereas the average correlation between the 18 lobar measures (six lobar measures by three variables including volume, thickness, and surface area) was 0.4. These strong correlations between measures indicate that Bonferroni corrections for the number of comparisons would be overly conservative. For this reason, the p-value adjustment was selected to take into account the correlation between measures (Sankoh et al., 1997), and a p-value threshold of 0.008 was used in the brain segment structures, ventricles, and ICV analyses, and a threshold of 0.009 was used in the lobar analyses. The presented tables report unadjusted p-values. Confidence intervals reported in the text are based on adjusted p-values.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

Morphometric differences between epilepsy and control groups

The difference in brain volume between subjects with epilepsy and healthy controls for each region of interest are shown in Tables 2 and 3. Each difference is expressed in standard deviation (SD) units to facilitate comparisons across regions, and all group differences are the result after accounting for differences in age between the two groups. Full models including results for both age and the effect of epilepsy are available from the authors.

Table 2.   Adjusted and unadjusted volume differences between epilepsy and control groups for regions of interest (raw values available from authors upon request)
 Unadjusted volumeHeight/gender adjusted
βStandard errorp-valueβStandard errorp-value
ICV−0.590.180.002−0.520.190.007
Brain segment      
 Amygdala−0.040.190.8460.060.200.777
 Brainstem−0.440.190.023−0.390.190.042
 Caudate−0.810.170.000−0.770.170.000
 Cerebellum cortex−0.580.170.001−0.490.170.004
 Cerebellum white−0.520.000.007−0.450.190.021
 Cerebral cortex−0.400.160.011−0.390.150.011
 Cerebral white−0.740.180.000−0.680.190.000
 Hippocampus−0.820.170.000−0.770.180.000
 Pallidum−0.500.180.007−0.430.180.021
 Putamen−0.180.180.313−0.190.180.289
 Thalamus−0.780.170.000−0.740.170.000
Ventricles      
 Lateral−0.140.190.469−0.020.190.924
 Inferior lateral0.090.190.6370.170.190.380
 Third−0.150.190.4240.000.190.995
 Fourth0.040.190.8530.100.190.616
Table 3.   Adjusted and unadjusted results for lobar volumes, thickness, and area between epilepsy and control groups (raw values available from authors upon request)
 Unadjusted volumeHeight/gender adjusted
βStandard errorp-valueβStandard errorp-value
TBV−0.500.160.003−0.450.160.006
Volume      
 Frontal−0.400.160.015−0.340.150.028
 Parietal−0.540.170.002−0.500.170.004
 Medial temporal−0.490.190.011−0.390.190.043
 Lateral temporal−0.660.170.000−0.630.160.000
 Occipital−0.260.190.170−0.200.190.287
 Cingulate−0.330.190.082−0.230.180.213
Area      
 Frontal−0.400.190.039   
 Parietal−0.370.190.056   
 Medial temporal−0.400.190.038   
 Lateral temporal−0.590.180.002   
 Occipital−0.070.190.726   
 Cingulate−0.320.190.098   
Thickness      
 Frontal−0.180.140.212   
 Parietal−0.460.160.004   
 Medial temporal−0.150.190.427   
 Lateral temporal−0.290.170.083   
 Occipital−0.240.180.194   
 Cingulate0.010.170.407   

Table 2 demonstrates distributed atrophy in the epilepsy subjects relative to healthy controls. When the raw volumes are evaluated, ICV is nearly 0.6 SD smaller (−0.59, 95% confidence interval [CI] −1.09 to −0.09) among those with epilepsy than among healthy controls. The highest levels of atrophy were found in the hippocampus (−0.82, 95% CI −1.29 to −0.35), caudate (−0.81, 95% CI −1.28 to −0.34), thalamus (−0.78, 95% CI −1.23 to −0.32), and cerebral white matter (−0.74, 95% CI −1.23 to −0.26). Significant differences were also found in the cerebellum cortex (−0.58, 95% CI −1.06 to −0.11), cerebellum white matter (−0.52, 95% CI −1.02 to −0.01), and pallidum (−0.50, 95% CI −1.00 to −0.01). Smaller differences were present in the brainstem (−0.52, 95% CI −0.95 to 0.08) and cerebral cortex (−0.40, 95% CI −0.83 to 0.02), but these differences did not reach statistical significance after adjusting for multiple comparisons. There is no evidence that the volume of the amygdala, putamen, or ventricles differed between the two groups.

Although height and gender explained up to 16% of the volume differences between the epilepsy and healthy controls, most of the atrophy observed in the raw values persisted after accounting for differences in height and gender. ICV remained more than 0.5 SD smaller (−0.52, 95% CI −1.03 to −0.01) among epilepsy subjects. The relative atrophy among epilepsy subjects was greatest within the hippocampus (−0.77, 95% CI −1.25 to −0.30), caudate (−0.77, 95% CI −1.24 to −0.29), thalamus (−0.74, 95% CI −1.19 to −0.29), cerebral white matter (−0.68, 95% CI −1.18 to −0.18), and the cerebellum cortex (−0.49, 95% CI −0.94 to −0.03). Smaller differences were also evident in the cerebellum white matter (−0.45, 95% CI −0.96 to 0.07), pallidum (−0.43, 95% CI −0.93 to 0.07), cerebral cortex (−0.39, 95% CI −0.81 to 0.02), and brainstem (−0.39, 95% CI −0.91 to 0.13), but none of these differences reached statistical significance after adjusting for multiple comparisons. There were no significant differences between groups for the amygdala, putamen, and ventricles.

The differences between the epilepsy and control groups (expressed as the number of SDs) within each of the six lobar regions as well as in the TBV can be found in Table 3. The generalized atrophy among those with epilepsy suggested by the brain segment analyses was confirmed in the analysis of TBV. The TBV was 0.5 SD (−0.50, 95% CI −0.96 to −0.05) lower among epilepsy subjects when raw volumes are compared and 0.45 SD (−0.45, 95% CI −0.89 to −0.01) lower when adjusted volumes are compared. When the lobar raw volume measures were compared between groups, the epilepsy group showed general atrophy across all lobar regions, although only the differences in the parietal and lateral temporal lobes reached statistical significance. The biggest difference occurred within the lateral temporal lobe, where the raw volume for patients with epilepsy was on average, two thirds of an SD (−0.66, 95% CI −1.11 to −0.20) lower than that for the healthy controls, although the parietal lobe was slightly more than a half a SD smaller in those with epilepsy (−0.54, 95% CI −1.00 to −0.07). Smaller differences were present in the frontal (−0.40, 95% CI −0.83 to 0.04) and medial temporal lobes (−0.49, 95% CI −1.00 to 0.02), but neither of these differences reached statistical significance after adjusting for multiple comparisons.

These differences were preserved once volumes were adjusted for differences in height and weight. After adjustment, the difference in volume between epilepsy and health control subjects remained statistically significant in the lateral temporal lobe (−0.63, 95% CI −1.07 to −0.19) and parietal lobes (−0.50, 95% CI −0.95 to −0.05). The differences between the epilepsy and healthy control subjects were further reduced in the frontal (−0.34, 95% CI −0.76 to 0.08) and medial temporal lobes (−0.39, 95% CI −0.90 to 0.13), but differences did not reach statistical significance after adjusting for multiple comparisons.

Comparison of lobar surface area and thickness measures between the epilepsy and healthy control subjects suggests that the group differences in volume within the lateral temporal lobe are due more to the smaller surface area among those with epilepsy, whereas differences in the parietal lobe are due to reduced thickness. Although all six lobes show smaller surface area among those with epilepsy, only the difference within the lateral temporal lobe (−0.61, 95% CI −1.09 to −0.08) reaches statistical significance. In addition, participants with epilepsy have less surface area than healthy controls in their frontal (−0.39, 95% CI −0.91 to 0.12) and medial temporal lobes (−0.39, 95% CI −0.91 to 0.12), but these differences did not reach statistical significance after adjusting for multiple comparisons. In contrast, differences between groups in lobar thickness were evident only in the parietal lobe in epilepsy subjects (0.61, 95% CI −0.89 to −0.04).

Age-related changes in control and epilepsy groups

Cross-sectional characterization of changes in brain structure as a function of chronologic age are depicted in Figs. 1 and 2. We tested different relationships between age and volume, including linear, quadratic, and natural log. Because these models compare the functional form of a variable rather than statistical significance, they are not nested models. For this reason, the appropriate function for age was determined based on the model fit statistic Akaike information criterion (AIC), which does not require models to be nested. There were no significant group by age interactions for the areas of interest depicted in Figs 1 and 2. For ventricle volumes, there were significant group by age interactions; these are presented in Fig. 3.

image

Figure 1.   Patterns of age-related brain changes for regions of interest in controls and participants with epilepsy. Blue, epilepsy group; orange, control group.

Download figure to PowerPoint

image

Figure 2.   Patterns of age-related brain changes for measures of lobar volume, thickness, and area in controls and participants with epilepsy. Blue, epilepsy group; orange, control group.

Download figure to PowerPoint

image

Figure 3.   Significant interactions between group and age for ventricular volumes. Blue, epilepsy group; orange, control group.

Download figure to PowerPoint

Figure 1 shows that the volume of the cerebral cortex and cerebral white matter exhibited a quadratic relationship with age. In the cerebral cortex, volume decreased until around age 50 after which it leveled out. In the cerebral white matter, this relationship was inverted; volumes increased until age 40 before declining slightly at older ages. The natural log of age provided the best fit for the hippocampus and cerebellum cortex, suggesting larger changes at younger ages than at older ages. Volumes declined with age in the hippocampus and cerebellum cortex. Volumes of the caudate, putamen, pallidum, and thalamus declined linearly with age. Volumes of the lateral ventricle, inferior lateral ventricle, and third and fourth ventricles increased linearly with age (not shown).

Figure 2 shows that within the lobar regions, both the frontal and lateral temporal lobes exhibited a quadratic relationship with age. In these regions volume declines until age 50, after which it changed little. The natural log of age provided the best fit for the parietal, occipital, and cingulate regions, which all showed decreases with age. In each case, these declines are larger at younger ages than at older ages. The volume of the medial temporal lobe decreases linearly with age.

Differences in age-related trajectories between epilepsy and control subjects were seen in several regions, although they did not reach statistical significance after adjusting for multiple comparisons. Figure 3 shows that for the lateral and third ventricles, brain aging differed between epilepsy and control subjects (lateral ventricle: 0.024, 95% CI −0.009 to 0.080; third ventricle: 0.021, 95% CI −0.002 to 0.083). In these ventricles there is no significant increase in volume among the controls. In contrast, the subjects with epilepsy exhibited a strong increase with age, such that after age 35 their ventricles are larger than those of healthy controls.

Examination of lobar parcellations for cortical analyses may be less sensitive than vertex analyses, which in some studies have revealed progression effects. Figure 4 depicts false discovery rate (FDR) corrected results, smoothed with a 25 mm kernel to improve intersubject averaging, depicting strong significant negative relationships between age and cortical thickness in the controls (top) and the TLE group (middle). Clearly, both groups show major age-related changes. The bottom panel shows the noncorrected p < 0.05 group differences with age, none of which survived corrections for multiple comparisons, results that are therefore consistent with the previously summarized lobar analyses.

image

Figure 4.   FDR corrected change in cortical thickness with increasing chronologic age in controls (top panel) and TLE patients (middle panel). The bottom panel shows uncorrected differences between groups–none survived correction.

Download figure to PowerPoint

The question arises whether laterality of seizure onset may exert variable brain aging profiles. A subset of the TLE subjects underwent ictal monitoring with confirmation of side of recurrent seizure onset (n = 13 left and 13 right TLE). We examined cortical thickness through vertex analyses in left TLE (LTLE) versus right TLE (RTLE) groups at baseline, as well as age regressions with cortical thickness within the LTLE and RTLE groups compared to controls. At baseline there were no age and FDR corrected differences between the LTLE and RTLE groups and there were no significant corrected age regression differences between controls and the RTLE and LTLE groups that would imply that age effects were accelerated within a lateralized group.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

Three core findings emerge from this investigation. First, the nature and range of neuroanatomic abnormalities in persons with mean childhood/adolescent onset TLE are extensive. They include the hippocampus as expected, but also extrahippocampal temporal lobe regions, diverse subcortical structures, cerebellum, brainstem, and extratemporal lobe gray and white matter. Second, age-related changes are evident for many of these anatomic structures and regions, with the pattern of age-trajectories largely comparable between the groups, with the epilepsy group always showing less volume and/or thickness with aging. Third, the primary exception to the above is age-accelerated expansion of the ventricular system (third and lateral ventricles) in the epilepsy participants.

Anatomic abnormalities in temporal lobe epilepsy

The TLE group exhibited abnormal (decreased) volumes of hippocampus, caudate, thalamus, cerebral white and gray matter, cerebellar white and gray matter, pallidum, and brainstem. Between-group differences in the amygdala and putamen volumes were not statistically significant. Regarding cortical regions, the TLE group exhibited smaller volumes of the lateral temporal, parietal, medial temporal, and frontal regions, but not occipital lobe or cingulate region. The total anatomic burden associated with this localization-related form of epilepsy is therefore notable and consistent with examinations of the anatomic consequences of early onset TLE in quantitative MRI investigations of prevalent cases (Sisodiya et al., 1997; Bernasconi et al., 2004; Mueller et al., 2006; Lin et al., 2007; Keller & Roberts, 2008; McDonald et al., 2008a,b,c; Hermann et al., 2009; Bonilha et al., 2010a,b; Li et al., 2012). The general breadth of abnormality detected here is notable for the mean age of the sample (approximately 35 years). Because these are patients with average childhood/adolescent onset of recurrent seizures, these findings infer a significant adverse neurodevelopmental effect on diverse brain structures. (Hermann et al., 2002a,b; Weber et al., 2007).

As expected, given previous published cross-sectional research in healthy populations (e.g., Pfefferbaum et al., 1994; Blatter et al., 1995; Toga et al., 2006; Sowell et al., 2007; Raz et al., 2010; Walhovd et al., 2011), advancing chronologic age was associated with volumetric reductions in healthy controls. In the cerebral cortex, volume decreased until around age 50 after which it plateaued. This relationship was inverted for cerebral white matter where volume increased until age 40 before declining slightly at older ages. Volumes declined with age in the hippocampus and cerebellum cortex, whereas linear declines were seen in the volumes of the caudate, putamen, pallidum, and thalamus, with linear increases with age in the lateral ventricle, inferior lateral ventricle, and third ventricles. Within the lobar regions, both the frontal and lateral temporal lobes revealed a quadratic relationship with age in which volume declined until age 50, after which there was little change. The natural log of age provided the best fit for the parietal, occipital, and cingulate regions, which all showed decreases with age. In each case, these declines were larger at young than at older ages. The volume of the medial temporal lobe decreased linearly with age.

Of interest, there were few interaction effects of the type that would imply that within the age range investigated the participants with epilepsy exhibited accelerated brain aging effects compared with healthy controls. Such effects were observed only in the lateral and third ventricles, and in these areas patients with epilepsy exhibited an accelerated age effect. There were no significant age accelerated findings for the patients with epilepsy in the remaining regions. Hence, brain aging effects appeared largely comparable for the two groups, with the participants with epilepsy showing significantly reduced volumes compared with controls across a wide range of anatomic areas at any age. The primary exception was the volumes of the ventricles (lateral and third but not fourth ventricle), which showed greater age-related expansion in the epilepsy group compared to the controls.

Abnormal expansion of the ventricular system has been observed in many neurologic and psychiatric disorders, including dementia (Carmichael et al., 2007), multiple sclerosis (Dalton et al., 2006), and autism spectrum disorder (Palmen et al., 2005). The natural history of ventricular enlargement and its consequences have been increasingly investigated and in some disorders, such as schizophrenia, enlargement is present early in the course of the disorder (Sowell et al., 2000; Vita et al., 2006) and progresses over time (Kempton et al., 2010). Ventricular volume has been studied closely in normal and abnormal aging (e.g., Carmichael et al., 2007; Carlson et al., 2008; Hua et al., 2008; Fjell et al., 2009; Chou et al., 2010), and although expansion of the lateral, inferior lateral, and third ventricles is observed in normal aging (Fjell et al., 2009), age-accelerated ventricular expansion has been shown to be predictive of conversion from normal cognition to mild cognitive impairment 2 years later (Carlson et al., 2008; Weiner, 2008), as well as a precursor of dementia and its course (Carmichael et al., 2007a,b; Nestor et al., 2008; Driscoll et al., 2009). Current work has suggested that accelerated ventricular expansion may serve as a biomarker not only of incipient dementing disorders given its ability to predict changes (declines) in cognition and memory, but to predict clinical ratings of functional status as well (e.g., Chou et al., 2010). Although we suggest caution given the cross-sectional nature of our investigation, we have also found lateral and third ventricular volume to expand to a significantly greater degree than controls over a prospective 4-year test–retest interval (Tuchscherer et al., 2010). Monitoring ventricular volume and associated prospective changes in cognition may be an informative task for the future, especially in an older cohort of subjects.

Lastly, the cumulative anatomic abnormality reflected in Tables 2 and 3 and Figs. 1 and 2 is increasingly appreciated (Blanc et al., 2011; Engel & Thompson, 2012), consistent with neuropsychological evidence (Oyegbile et al., 2004), painting a picture of considerable neurobiologic abnormality that continues apace with increasing chronologic age. The outer limit of our sample is 60 years of age and distant from the point where traditional neurodegenerative disorders begin to appear. How these subjects fare with increasing chronologic age with this neuroanatomic profile would seem to be an important issue.

Limitations

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

This investigation has several limitations that should be recognized. First, this is a cross-sectional study, the limitations of which are appreciated. This investigation is an attempt to comprehensively characterize the lifespan trajectory of diverse brain structures in chronic TLE, as has been done in studies of cognition, in order to determine whether the “course” of identified brain abnormalities differs compared to controls. Clearly a prospective investigation is the preferred approach, but it may be unlikely for a lifespan study of this type ever to be accomplished. This approach is novel and may spur others to undertake similar analyses where the reliability of these findings will be determined and with larger samples of patients with lateralized TLE to determine if this is a pertinent consideration in brain aging.

Second, we examined total volumes of lateralized (left and right hemisphere) structures. One of the strengths of processing systems such as FreeSurfer is that it provides data about diverse anatomic structures, thereby enabling characterization of the broader landscape of brain structure and change compared to traditional region of interest approaches. Associated with this advantage, however, is considerable output and increased risk of type 1 error. A much larger sample would be required to examine lateralized anatomic findings in persons with confirmed left and right TLE to determine whether ipsilateral anatomic would show a different pattern of brain aging compared to contralateral structures.

Third, another concern in cross-sectional research of this type is a cohort effect. Older subjects may be at disadvantage due to histories of poorer health, increased exposure to older and more harmful medications, or other generational effects. But here we see no convincing consistent evidence of cohort effects across our measures of interest.

Fourth, the number of subjects investigated is modest and the outer limits of age remain comparatively young. It is possible that the trajectories presented here may change (i.e., become more abnormal) with inclusion of subjects of older chronologic age in whom increasingly adverse neuroanatomic changes are known to occur.

Finally, modal cognitive and structural profiles are helpful in that they provide an overall characterization of disease effect. However, considerable individual variability is typically hidden within these modal profiles—a point we have discussed and demonstrated previously (e.g., Dabbs et al., 2009). We suspect that similar variability exists with respect to neuroanatomic measures and that distinct phenotypes may exist, including subsets that may be comparable to controls but others that may exhibit more progressive changes with time. These types of analyses remain to be done.

Disclosures

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References

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.’

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Disclosures
  8. References