Hippocampal volume in children with temporal lobe epilepsy compared to healthy children : A magnetic resonance imaging study
Context : There are currently few published studies that compare hippocampal volume (HCV) in children with temporal lobe epilepsy (TLE). Aims : T0 o compare HCVs in children with TLE and in relation to normal controls (NC), and to analyze HCV change in the acute phase in pediatric subjects using magnetic resonance imaging (MRI). Setting and Design : T0 he case group (n=24) was matched in gender and age range with NC subjects (n=24). Subjects were divided into three groups according to age: 2-5.9 years, 6-8.9 years and 9-13 years. Materials and Methods : M0 anual measurements were used to obtain HCVs on oblique coronary MRI images obtained by three-dimensional magnetization-prepared rapid gradient-echo (3D-MPRAGE) sequence. Statistical Analyses : T0 he HCV was calculated and normalized. Paired t-tests were used to compare the right and left HCVs within each study group. Analysis of covariance (ANCOVA) was used to consider the impact of age and gender on HCVs. Results : T0 wenty-two patients were diagnosed with definite/probable TLE, and two were non-syndromic focal epilepsy which was distributed in the 2-5.9 years group. The range of left to right mean HCVs of the case and NC group aged 2-13 years was 2014.14 ± 54.32 mm 3 to 2165.31 ± 80.99 mm 3 and 2015.46 ± 26.97 mm 3 to 2100.93 ± 57.33 mm 3 respectively. There were age-related differences in left and right HCV, but no effect of gender. Relative to NC subjects, cases group aged 2-5.9 years had significantly different HCVs, while no significant difference was found in the other two groups. There was significant difference in the right to left HCVs in the case subjects aged 2-5.9 years, but not in the other age groups. Conclusions : T0 he heterogeneity in the 2-5.9 years age cohort may relate to the increased HCVs. The HCV data from NC subjects may be used as a reference to assess hippocampal abnormalities in clinical practice.
Keywords: Hippocampal volume, hippocampus, magnetic resonance imaging, temporal lobe epilepsy
Abnormalities in the hippocampus (HC) can be examined using a variety of existing methodologies. Of these, magnetic resonance imaging (MRI) provides an objective means for studying the individual temporal lobe structures in vivo. This has been commonly used to identify patients with temporal lobe epilepsy (TLE) and identify hippocampal sclerosis (HS). Unilateral HS is the most common pathological diagnosis in TLE, and up to 65% of cases of TLE may be attributed to lesions arising exclusively in the HC.  MR-based hippocampal volume (HCV) measurement offers a practical and valuable method for both clinical and research purposes. This quantitative assessment may be helpful in establishing early and accurate diagnosis, measuring disease progression, detecting and assessing therapeutic effects and predicting prognosis in different diseases. ,
We conducted this study and used a manual tracing technique primarily to establish the volumetric range of HC in children with TLE or non-syndromic partial epilepsy and in normal children within a much broader spectrum of age (between 2 and 13 years old) than previously reported using MRI. Clinically, TLE is not always recognized at initial onset, and HS is rarely seen in new-onset patients. ,, Our second aim was to explore the HCV change in pediatric TLE subjects in relation to normal controls. The case group children in this study were diagnosed by clinical history and electroencephalogram (EEG) examination, manifested with normal routine MRI and in the absence of history of febrile seizures or patients with mesial temporal lobe epilepsy and HS (MTLE-HS).
The case group comprised 24 children. Twenty-three children had manifested complex partial seizures (CPS) with secondary generalization with interictal data localized to one or both temporal lobes and one child manifested simple partial seizure with interictal data localized to the right temporal lobe. Seizures were classified according to the International League Against Epilepsy (ILAE) classifications (Commission on Classification, 1981; Commission on Classification, 1989). , The cohort was defined as definite TLE based on a presence of complex partial seizures (with or without aura) of temporal semiology with localized temporal epileptiform abnormalities (n=18), probable TLE as complex partial seizures (with or without aura) of uncertain semiology with localized temporal interictal epileptiform discharges (IEDs) (n=4); and non-syndromic focal epilepsy if the electroclinical syndrome could not be defined despite the suggestion of a temporal semiology (n=2) according to the criteria defined by the ILAE (1989) [Table 1]. All case group children were divided into the above groups based on the type of epilepsies with seizure descriptions over two times and routine results of scalp EEG or Video-EEG monitoring (VEM). There were 14 boys and 10 girls with the mean age of 7.29 ± 2.88 years (range = 2-12.6 years). The criteria for inclusion in the study were (1) definite/probable TLE or non-syndromic focal epilepsy with possible temporal localization; (2) epileptic activity localized in temporal lobe by EEG or VEM examination; (3) no other neurological disorder; (4) normal conventional MRI findings. All case children were without febrile seizures and MTLE-HS, children with a positive family history of febrile seizures or familial MTLE were excluded. Other exclusion criteria were cerebrovascular tumor or traumatic diseases, febrile seizures or a history of encephalitis, meningitis, or mental retardation, a history of hypoxia or hypoglycemia, and psychiatric illness or cognitive impairment. All research scans had been reviewed by two radiologists who had visually assessed the HC formations for atrophy and signal change and the entire brain for other abnormalities including signs of subtle focal cortical dysplasia. MRI criteria of HS included volume loss, loss of architecture (digitations) and increased hippocampal signal.
Twenty-four control participants (NC group) were selected to ensure a match in gender distribution and to reflect the age range of the patient case group at the time of MRI scan. The mean age of control participants was 7.28 ± 2.70 years (range = 2.5-12.6 years). Inclusion criteria were a normal neurological examination and an absence of cerebrovascular, tumor and traumatic disease. Controls with a history of febrile seizures or positive family history of epilepsy were excluded. The study had the local ethics committee approval and all subjects or parents gave their informed consent for participation.
In view of the age distribution of the cohort (between 2 and 13 years old) in this study and the changes in HCV with age in children, the 24 case subjects were arbitrarily divided into three groups according to age with a difference of 3-4 years: 2-5.9 years (n=8); 6-8.9 years (n=10); 9-13 years (n=6). The same method was used in the NC group.
Magnetic resonance imaging methods
MRI analyses were performed on a GE 3.0 Tesla Signa HDxt Unit (GE Medical Systems, Milwaukee, WI, USA) and workstation. After the symmetrical positioning of the head on axial and sagittal images, we obtained contiguous 1-mm-thick sections using three-dimensional magnetization-prepared rapid gradient-echo (3D-MPRAGE) imaging. HCVs were measured from the oblique coronal MR images perpendicular to the long axis of HC. MRI sequences and parameters are presented in [Table 2]. We could not perform T2-relaxometry on the hippocampal regions due to technical reasons and relied on T2-FLAIR images to define the signal abnormalities. The total acquisition time was 6 min and 30 sec for each subject.
Hippocampal volume measurement
Boundaries of the HC were determined according to the method described by Watson and Jack et al.,,,, HCVs were manually delineated on successive coronal slices using a modified protocol based on previously published methods. ,,, The left and right parts of HCV were obtained using manual tracing [Figure 1] and [Figure 2]a-c. A slice volume was calculated by multiplying the area outlined by the slice thickness. The whole volume was calculated by adding all the slice volumes. The total average time required for a trained individual to segment and calculate the HCV was 1 h. The trained individual was blinded to the clinical data of all subjects.
The total intracranial volume (ICV) measurement was used to correct for intersubject variation in head size. Total ICV was measured by multiplying anterior to posterior (A-P), up to down (U-D), left to right (L-R) distance of the head. A-P line was the distance from frontal to parietal of the cranial inner plate from the middle sagittal image. U-D line was the distance from the anterior edge of the foramen magnum to the top inner plate and perpendicular to the A-P line in the same image. L-R line was the maximum diameter of the dorsal thalamus plane from the axial image.
Normalization of pediatric HCV was required to allow absolute volume comparison between various ages.  This method described by Jack et al., , derives a corrected HCV using the following equation : N0 V = OV − Grad (CMi − CM mean ). NV is normalized HCV, OV is original HCV, Grad is the gradient of the regression line between the HCV and the total ICV measure, CMi is the value of the appropriate ICV measurement for that subject, and CM mean is the mean value of that measure for all NC subjects.
HCVs were normalized to control variations in ICV using analysis of covariance (ANCOVA). The raw and normalized volume samples were checked using normal distribution tests and expressed as mean ± standard deviation (SD). The mean value and 95% confidence intervals were calculated for each age and gender group separately for each side. Paired t-tests were used to compare the difference between the right and left HCVs within each group. ANCOVA was used to consider the impact of age and gender on HCVs. All statistical analyses were calculated with SAS statistical software (Version 8.2). The level of statistical significance was P < 0.05 (two-tailed).
Response to antiepileptic drug
Twenty-three (95.83%) patients received antiepileptic drugs (AEDs) for not more than three years. Two definite TLE patients in the 2-5.9 years group, three patients (two definite TLE and one probable TLE) in the 6-8.9 years group and one probable TLE patient in the 9-13 years group continues to have seizures after receiving AEDs over two months. The other 17 patients were seizure-free after AEDs over two months and all of them are following up.
Hippocampal volumes before normalization
The mean left and right HCVs before normalization were 2050.40 ± 70.04 mm 3 and 2073.93 ± 62.15 mm 3 among case children and 2053.65 ± 73.65 mm 3 and 2075.59 ± 77.59 mm 3 among NC children respectively.
Hippocampal volumes after normalization
The normalized mean HCV in the left and right was 2075.54 ± 74.60 mm 3 and 2099.06 ± 80.64 mm 3 in case children, 2053.65 ± 54.40 mm 3 and 2075.59 ± 68.22 mm 3 in NC children. Both sides of HCV increased by age before normalization, and decreased by age after normalization. The normalized HCV of the three age groups are presented in [Table 3] and [Figure 3] and [Figure 4]. Normal distribution tests and 95% confidence intervals of left and right HCV among NC subjects were 2030.68 to 2076.63 mm 3 and 2046.78 to 2104.39 mm 3 respectively.
Effect of age and gender on hippocampal volumes
Statistically significant differences were found between HCV and age (P < 0.0l). In the left side of the cases group t=8.06, F=65, P < 0.001 and in the right side of cases group t=8.03, F=64.47, P < 0.001. In the left side of NC group t=6.87, F=47.22, P < 0.001 and in the right side of NC group t = 6.05, F = 36.56, P < 0.001. There were no statistically significant differences in either raw or normalized volumes between gender groups (P>0.05).
The symmetry analysis of the hippocampus
The right-side HC was consistently larger than the left in both the cases group and NC group (raw and normalized volume). The right normalized HCV was larger than the left side (mean value 18.09 ± 65.87 mm 3 ), although the difference was not statistically significant (P=0.1915, P>0.05).
T1-weighted image 3D-MPRAGE was used to improve the contrast between the HC and the surrounding structure. Image quality and contrast between gray and white matter are superior with the 3D-MPRAGE sequence compared with the T1-weighted spin echo (SE) sequence. , In this study we could obtain contiguous thin-section (1 mm) images with readily acquired multiplanar reformations obviating additional images and could depict more focal lesions. The 3D-MPRAGE sequence had 228-256 slices of which 60 slices on average contained the pediatric HC in raw data in this study. Although a thinner slice may be advisable after reformation, Laakso and colleagues have shown that acquired HCVs using 1, 3 or 5 mm slice thickness do not differ significantly.  For reasons of time and clinical practicality, contiguous 3-mm-thick sections were reformatted and measured in this study.
Manual tracing technique was used in this study because it provides a more precise technique compared to automatic methods and it is believed that this will continue to remain the "gold standard" in future research. ,,,, As the pediatric HC is still developing, it is smaller and less well-defined than that observed in adults and thus both automatic and manual segmentation are more difficult. There are currently few studies which compare pediatric HCVs. A retrospective study of pediatric HCVs reported normative volumes for HC ranging from 1.04 to 3.47 mL.  Obenaus et al., found the mean HCV of the left and right side to be 2.44 mL and 2.59 mL in six NC children aged 14-60 months.  No statistically significant differences were found between the two sides, and four of the subjects showed 10% larger right HCV than left side. Mulani et al., found the mean HCV of the left and right side to be 2.49 cm 3 and 2.75 cm 3 in 20 children aged 6 to 12 years.  In the current study, the mean HCV was 2.053 cm 3 and 2.076 cm 3 for the left and right respectively in 24 NC children ranging in age of 2.5 to 12.6 years. These differences in the reported pediatric HCVs are likely due to sample heterogeneity, differences in regionality and ethnicity, MRI acquisition sequences, post-processing of the data, methods used in delineation of the HC, normalization methods, and operator proficiency.
Brain volume correlates with age, with each increasing over time. We found statistically significant differences among HCV and age groups in this study. This result is in agreement with some previous studies , but not with others. , This study confirmed previous reports showing significantly greater right HCVs than left HCVs in children. , This right-left asymmetry has also been observed in adult HCV. ,
Among the few childhood epilepsy morphometric studies with positive findings, Lawson et al., reported HCV reduction and HC asymmetry in MTLE with mean duration of 7.9 years. , Studies of newly diagnosed epilepsy typically fail to find many patients with clear HS at onset. While there is evidence from adult studies and studies in chronic epilepsy patients that HC atrophy may be a progressive lesion, there is little information regarding HC abnormalities early in the course of epilepsy in patients, particularly in children, Classic MTLE with HS is an uncommon finding in the general population.  Patient loss in HCV has not been consistently related to longer duration of illness or earlier age of onset.  In this study we have not found HCV reduction in our diagnosed definite/probable TLE children.
Total HCV increased in the 2-5.9 years' case group compared with that of NC in this study. Such results were not found in the 6-8.9 or 9-13 years' age groups. SD of the right and left HCV in the case cohort was wider than in the NC cohort, especially in the 2-5.9 years' case group. The difference in the SD values could be responsible for the P value differences between the groups. There was more greater proportion of cases of definite TLE (n=3) and probable TLE (n=2) in the right side of the 2-5.9 years age group than the other two groups.This may cause the significant differences comparison between LHCV to RHCV in the 2-5.9 years group. Retrospective analysis of the clinical data as shown in [Table 1] indicates that there is no significant difference in the mean interval between seizures and MRI in days. Comparison of definite/probable TLE between the three groups [Table 4] showed a statistically significant result in the 2-5.9 years age group. Heterogeneity of the case cohort in differences in the clinical profile of the children in the 2-5.9 years' age group may be related to the significant difference in the results between the three groups. Duration of seizure lasted from a few seconds to 15 mins in the 2-5.9 years' age group, which was longer than in the 6-8.9 (a few seconds to 5 min) and 9-13 years' age groups (a few seconds to 3 min). Longer duration of seizures and increased frequency of seizure episodes may be related to cytotoxic edema and may cause increased excitability of the HC and HCV increases in the acute phase of temporal IEDs. Completing MRI examination as soon as possible after onset of epilepsy may lead to the observation of HCV increases.
Few studies have reported HCV change in the acute phase of TLE in a pediatric population. The study of Berg et al., found the significance of bilaterally large or small hippocampus is unclear.  Both large and small extreme volumes were modestly associated with a clinical assessment of whether the individual epilepsy was likely to be TLE. The results from several studies suggest that HC anomalies are not limited to patients with clear unilateral TLE. ,,, The heterogeneity of the 2-5.9 years' group with definite TLE (n=4), probable TLE (n=2) and non-syndromic focal epilepsies (n=2) may have caused the significant result. There are some limitations in the present study that need to be highlighted. The timing of performance of MRI scan was non-uniform and this is a major drawback for MRI volmetry study. The diagnosis of TLE/probable TLE was based predominantly on seizure descriptions and routine EEG, VEM being obtained in only three patients in the case cohort. Another limitation of the present study which should be noted is that the sample size is relatively small, and large sample sizes of exclusive TLE cohorts' researches with long-term prospective follow- up assessments are needed to confirm our findings. MRI acquisition sequences, post-processing of the data, methods used in delineation of the HC, normalization methods, operator proficiency and more importantly the timing of the scan was in relation to the seizure.
There were two main findings in this study. First, the mean HCV in the left and right side in the current study was 2.075 cm 3 and 2.099 cm 3 respectively in TLE or non-syndromic focal epilepsies' children, and the mean HCV in the left and right side hippocampus in NC children aged 2-13 years was 2.053 cm 3 and 2.075 cm 3 respectively after normalization. A correlation between age and HCV was observed but there was no correlation with gender. Second, the HCV of both sides increased in the 2-5.9 years case subjects compared with that of the NC group in this study. This variation may be a consequence of the inherent clinical differences between the age groups and the timing of performance of the MRI analysis and largescale homogenous studies are required to derive further information on HCV changes across a pediatric TLE cohort.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3], [Table 4]