Progressive Hippocampal Volume Reduction Beyond Normal Aging in Temporal Lobe Epilepsy: A Longitudinal MRI Study

Abstract

Recent neuroimaging studies have suggested that epilepsy may show progressive brain structural changes over time. However, longitudinal hippocampal volume changes in temporal lobe epilepsy (TLE) have not been adequately compared with those observed in normal aging. In this study, we enrolled 46 patients with unilateral TLE and 25 healthy controls with longitudinal structural MRI scans using the same scanner. Hippocampal volumes were quantified using the Hipposeg software, and the annual percentage volume reduction relative to baseline (%Reduction/year) was calculated for each hippocampus. Compared to healthy hippocampi (median −0.37% per year, IQR = 3.97), both ipsilateral (median −1.04% per year, IQR = 3.65) and contralateral (median −1.08% per year, IQR = 3.17) hippocampi in TLE showed significantly greater volume reduction over time (both p < 0.05). Furthermore, annual ipsilateral hippocampal volume reduction was significantly correlated with disease duration (Spearman's rs = 0.366, FDR‐corrected p = 0.036), suggesting that hippocampal atrophy progresses more rapidly during the early stage after epilepsy onset. These findings indicate that hippocampal volume loss in TLE exceeds that expected from normal aging and provide further evidence supporting progressive structural alterations during the course of the disease.

Plain Language Summary

Temporal lobe epilepsy may cause progressive changes in the brain beyond those expected from normal aging. Using repeated MRI scans, we found that the hippocampus showed greater volume loss over time in people with temporal lobe epilepsy than in healthy individuals. This volume loss occurred bilaterally and appeared to be greater during the earlier stages of the disease. These results provide further evidence supporting progressive structural brain changes in TLE and contribute to a better understanding of disease progression and abnormal brain aging in epilepsy.

Key points.

• We investigated longitudinal hippocampal volume changes in 46 patients with unilateral TLE and 25 healthy controls.

• We demonstrated that hippocampal volume loss in TLE exceeded that expected from normal aging in both sides.

• Furthermore, annual ipsilateral hippocampal volume loss was significantly associated with disease duration.

• The results suggested progressive hippocampal atrophy particularly during the early stage of the disease in TLE.

1. INTRODUCTION

Epilepsy is a common neurological disorder affecting approximately 50 million people worldwide, ¹ and temporal lobe epilepsy (TLE) is the most common focal epilepsy in adults. Patients with TLE experience a substantial disease burden not only because of recurrent seizures but also owing to a wide range of neuropsychiatric and cognitive complications. ² However, the neural mechanisms underlying the heterogeneous clinical course of TLE remain poorly understood.

Recent neuroimaging studies have suggested that epilepsy is associated with progressive structural brain changes over time. A longitudinal MRI study demonstrated progressive cortical thinning beyond that expected from normal aging in patients with focal epilepsy, with most affected regions being anatomically connected to the hippocampus. ³ Subsequently, the same group reported that epilepsy surgery may attenuate this progressive cortical thinning in TLE. ⁴ More recently, data‐driven machine learning approaches applied to cross‐sectional MRI data have identified distinct patterns of structural disease progression in epilepsy. ⁵ , ⁶ In contrast, longitudinal changes in subcortical structures, particularly the hippocampus, have not been sufficiently investigated. Given the pivotal role of the hippocampus in memory and emotional processing, ⁷ characterization of longitudinal hippocampal atrophy may provide important insights into the mechanisms underlying disease progression and the heterogeneous clinical manifestations of TLE. Because hippocampal sclerosis is the pathological hallmark of many cases of TLE, characterizing its longitudinal structural changes may provide a unique window into the biological processes underlying disease progression.

In the present study, we hypothesized that longitudinal hippocampal volume loss in patients with TLE exceeds that expected from normal aging. Using longitudinal structural MRI data acquired on the same scanner, we compared the annual rate of hippocampal volume reduction between patients with TLE and healthy controls. We further examined the associations between longitudinal hippocampal volume changes and clinical characteristics.

2. METHODS

2.1. Subjects

Longitudinal MRI data were obtained from the cohort of our previous cross‐sectional epilepsy study. ⁸ The detailed inclusion and exclusion criteria have been described previously. Briefly, the inclusion criteria were: (i) focal seizures consistent with TLE and (ii) unilateral temporal epileptiform discharges on routine scalp electroencephalography (EEG). The exclusion criteria were: (i) a history of acute encephalitis, meningitis, severe head trauma, hypoxic–ischemic encephalopathy, or brain surgery; and (ii) epileptogenic lesions on MRI other than unilateral hippocampal sclerosis (HS), such as brain tumors, vascular malformations, or destructive lesions.

Among the patients with TLE, we included those who had undergone two MRI examinations with an interval of at least 3 months. Healthy controls (HCs) who had also undergone two MRI examinations on the same scanner with an interval of at least 3 months were recruited for comparison. Ultimately, 46 patients with unilateral TLE and 25 HCs were included in the study (Table 1). All participants gave written informed consent. The study was approved by the Institutional Review Board at the National Center of Neurology and Psychiatry Hospital.

TABLE 1.

Demographics of patients with temporal lobe epilepsy and healthy controls.

	TLE (N = 46)	HC (N = 25)	p
Mean [SD] age at baseline (year)	44.5 [15.2]	49.2 [13.5]	0.205
Gender (M: F)	21: 25	10:15	0.646
Mean [SD] scan intervals (months)	23.9 [17.7]	19.6 [22.9]	0.385
Median [IQR] onset age	20 [28]	NA	NA
Median [IQR] duration at baseline	18.7 [27.8]	NA	NA
Focus side (R: L)	24:22	NA	NA
Hippocampal sclerosis (N)	19	NA	NA
Seizure free with ASMs (N)	6	NA	NA
Median [IQR] no. of ASMs at baseline	2 [2]	NA	NA

2.2. MRI acquisition and processing

Structural MRI was performed for all participants using the same 3.0‐T scanner (Philips Medical Systems, Best, the Netherlands) with the following parameters: repetition time (TR), 7.18 ms; echo time (TE), 3.46 ms; flip angle, 10; number of excitations (NEX), 1; in‐plane resolution, 0.68 × 0.68 mm²; slice thickness, 0.6 mm without an interslice gap; 300 slices; acquisition matrix, 384 × 384; and field of view (FOV), 26.1 × 26.1 cm. High‐resolution T2‐weighted and fluid‐attenuated inversion recovery (FLAIR) images were also acquired for visual assessment of potential epileptogenic lesions. Hippocampal volumes were quantified using the Hipposeg software (http://niftyweb.cs.ucl.ac.uk/program.php?p=HIPPOSEG), an automated segmentation tool specifically developed for accurate hippocampal delineation in patients with TLE as well as healthy controls and previously shown to achieve accuracy comparable to manual segmentation. ⁹ Hipposeg generated intracranial volume (ICV)‐corrected hippocampal volumes for each MRI scan. ICV showed excellent longitudinal reliability (intraclass correlation coefficient = 0.997), confirming that ICV remained highly stable across the two time points. The validity of the automated segmentation was visually confirmed for all participants, and a representative segmentation result is shown in Figure 1.

FIGURE 1.

Sample images of longitudinal MRI scans and hippocampal segmentation with volume evaluation by Hipposeg in a patient with TLE.

2.3. Calculation of volume changes per year

Longitudinal hippocampal volume change was calculated using the following equation:

$$\%\text{Reduction}=\frac{\text{Volume}\left(\text{followup}\right)-\text{Volume}\left(\text{baseline}\right)}{\text{Volume}\left(\text{baseline}\right)}$$

Here, %Reduction represents the percentage change in hippocampal volume relative to baseline, with more negative values indicating greater volume loss. The annual percentage volume reduction (%Reduction/year) was obtained by dividing %Reduction by the MRI follow‐up interval (years).

Annual hippocampal volume reduction was evaluated separately for (i) ipsilateral hippocampi in patients with TLE, (ii) contralateral hippocampi in patients with TLE, and (iii) healthy hippocampi (both right and left hippocampi in HCs).

2.4. Statistics

Statistical analyses were performed using SPSS Statistics version 25 (IBM SPSS, Tokyo, Japan). Demographic variables were compared between patients with TLE and HCs using unpaired t‐test for continuous variables and Pearson's χ ² test for categorical variables. Annual hippocampal volume reduction (%Reduction/year) was compared among ipsilateral, contralateral, and healthy hippocampi using the Kruskal–Wallis test followed by Dunn's post hoc test because normality was not satisfied according to the Shapiro–Wilk test. Associations between annual hippocampal volume reduction and baseline clinical variables, i.e., disease duration, age at seizure onset, and number of antiseizure medications (ASMs), were assessed using Spearman's rank correlation analysis with false discovery rate (FDR) correction for multiple comparisons. A two‐sided p value <0.05 was considered statistically significant.

3. RESULTS

3.1. Demographics

The demographic and clinical characteristics of the study participants are summarized in Table 1. Among the 46 patients with TLE, unilateral hippocampal sclerosis (HS) was identified in 19 (TLE‐HS), whereas the remaining patients showed no visually detectable epileptogenic lesions on MRI (TLE‐no). Most patients had drug‐resistant epilepsy despite antiseizure medication treatment, and the epileptogenic focus was almost evenly distributed between the right and left temporal lobes. There were no significant differences in age or sex between patients with TLE and HCs. In addition, the mean MRI follow‐up interval was comparable between the two groups (Table 1).

3.2. Longitudinal hippocampal volume reduction

Longitudinal hippocampal volume changes for each group and hemisphere are shown in Figure 2A. Compared with healthy hippocampi (median, −0.37%/year; IQR, 3.97), both ipsilateral (median, −1.04%/year; IQR, 3.65) and contralateral (median, −1.08%/year; IQR, 3.17) hippocampi in patients with TLE exhibited significantly greater annual volume reduction (Figure 2B).

FIGURE 2.

(A) Changes in the hippocampal volumes corrected for ICV in each category and laterality. (B) Hippocampal volume reductions per year across the ipsilateral, contralateral, and healthy hippocampi. (C) A significant correlation between ipsilateral hippocampal volume loss and disease duration. The dotted line represents a locally estimated scatterplot smoothing (LOESS) curve.

The median (IQR) annual hippocampal volume reduction was −0.79% (5.31) for ipsilateral hippocampi in TLE‐HS, −1.03% (1.91) for contralateral hippocampi in TLE‐HS, −1.44% (3.21) for ipsilateral hippocampi in TLE‐no, and −1.13% (3.20) for contralateral hippocampi in TLE‐no. There was no significant difference in the annual rate of ipsilateral hippocampal volume reduction between the TLE‐HS and TLE‐no groups (p = 0.415, Mann–Whitney U test).

3.3. Associations with clinical parameters

Annual ipsilateral hippocampal volume reduction was significantly correlated with disease duration (Spearman's rs = 0.366, FDR‐corrected p = 0.036; Figure 2C). This positive correlation indicates that greater hippocampal volume loss occurred in patients with a shorter disease duration. No significant associations were observed between annual hippocampal volume reduction and the other clinical variables or for the contralateral hippocampus.

4. DISCUSSION

In the present study, we investigated longitudinal hippocampal volume changes in patients with TLE using serial MRI acquired on the same scanner. We demonstrated that hippocampal volume loss exceeded that expected from normal aging in both the ipsilateral and contralateral hippocampi. Furthermore, annual ipsilateral hippocampal volume loss was significantly associated with disease duration, suggesting that hippocampal atrophy progresses more rapidly during the early stage of the disease.

Despite the longstanding debate regarding disease progression in epilepsy, relatively few longitudinal imaging studies have directly compared patients with healthy controls. ¹⁰ An earlier voxel‐based longitudinal MRI study demonstrated progressive gray matter loss in multiple brain regions in TLE, whereas no significant changes were observed in healthy controls ¹¹ ; however, that study did not include a direct statistical comparison between the two groups. More recently, a longitudinal MRI study demonstrated progressive cortical thinning beyond normal aging in focal epilepsy, with many of the affected cortical regions anatomically connected to the hippocampus. ³ Together with these previous findings, our results further support the concept that the hippocampus is one of the key structures involved in progressive brain changes associated with TLE.

An additional finding of the present study was the significant longitudinal volume reduction in the contralateral hippocampus. Structural abnormalities in epilepsy are increasingly recognized as extending beyond the epileptogenic focus through widespread network alterations. Therefore, progressive contralateral hippocampal atrophy may reflect network‐level pathological processes rather than localized structural damage alone, providing further evidence that TLE should be regarded as a network disorder.

Although the present study focused specifically on hippocampal volume changes, whether the hippocampus undergoes more rapid progressive atrophy than other brain regions remains an important question. Because Hipposeg is specifically designed for hippocampal segmentation, direct comparison of hippocampal atrophy with structural changes in other brain regions was beyond the scope of the present analysis. A previous longitudinal study reported mean cortical thinning rates of 0.024 mm/year in patients with epilepsy and 0.011 mm/year in healthy controls, ³ corresponding to an approximately twofold greater rate in patients with epilepsy. In the present study, the median annual hippocampal volume loss was 1.04%–1.08% in patients with TLE compared with 0.37% in HCs. Although these measures are not directly comparable because they represent different structural metrics, the magnitude of disease‐related change appears broadly similar. Future studies using comprehensive whole‐brain longitudinal analyses are needed to determine whether the hippocampus is particularly vulnerable to progressive atrophy in TLE.

We also observed a significant association between ipsilateral hippocampal volume loss and disease duration. A recent machine learning study using cross‐sectional MRI data identified a hippocampal‐predominant subtype of focal epilepsy, in which ipsilateral hippocampal atrophy appears early and is followed by progressive involvement of other subcortical and cortical regions, ⁵ and most TLE patients showed the hippocampal subtype pattern. Although derived from cross‐sectional data, these findings are consistent with our longitudinal observations and support that the ipsilateral hippocampal atrophy occurs in the early phase of disease course in TLE.

Several limitations should be acknowledged. First, the sample size was relatively small, and this was a single‐center study, which may limit the generalizability of our findings. Larger multicenter longitudinal studies are warranted to validate these results. Second, considerable interindividual variability was observed in longitudinal hippocampal volume changes, and some participants even demonstrated apparent volume increases during follow‐up. This variability likely reflects the inherent limitations of structural MRI‐based volumetric measurements as well as potential biological influences. Although none of the participants experienced a seizure immediately before MRI acquisition, peri‐ictal changes such as transient hippocampal swelling may still have influenced volumetric measurements. Third, age‐related structural brain changes are known to be nonlinear, ¹² whereas our analysis assumed a constant annual rate of volume change. Future studies incorporating nonlinear aging models may provide more accurate estimates of disease‐related atrophy. Finally, several potentially important clinical variables, including seizure frequency, were not available for analysis. The influence of these and other unmeasured confounding factors should therefore be considered when interpreting our findings.

5. CONCLUSION

Our findings demonstrate that longitudinal hippocampal volume loss in TLE exceeds that expected from normal aging and suggest that hippocampal atrophy is most pronounced during the early stage of the disease. These results provide further evidence supporting progressive structural brain changes in TLE and contribute to a better understanding of disease progression and abnormal brain aging in epilepsy.

AUTHOR CONTRIBUTIONS

Conceptualization: DS and NS; Formal Analysis: DS; Resources: DS, YS, RK, YK, HM, and NS; Writing – Original Draft Preparation: DS; Writing‐review & editing: YS, RK, YK, HM, and NS. All authors read and approved the final version of the manuscript.

FUNDING INFORMATION

This study was supported by the Japan Epilepsy Society: UCB Epilepsy Grant.

CONFLICT OF INTEREST STATEMENT

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.

ETHICS STATEMENT

This study adheres to the principles of the Declaration of Helsinki and was approved by the Institutional Review Board at the National Center of Neurology and Psychiatry Hospital (A2013‐039).

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.