Elsevier

NeuroImage

Volume 25, Issue 3, 15 April 2005, Pages 1016-1021
NeuroImage

Voxel-based morphometry of the thalamus in patients with refractory medial temporal lobe epilepsy

https://doi.org/10.1016/j.neuroimage.2004.11.050Get rights and content

Abstract

Previous research has suggested that patients with refractory medial temporal lobe epilepsy (MTLE) show gray matter atrophy both within the temporal lobes as well as in the thalamus. However, these studies have not distinguished between different nuclei within the thalamus. We examined whether thalamic atrophy correlates with the nuclei's connections to other regions in the limbic system. T1-weighted MRI scans were obtained from 49 neurologically healthy control subjects and 43 patients diagnosed with chronic refractory MTLE that was unilateral in origin (as measured by ictal EEG and hippocampal atrophy observed on MRI). Measurements of gray matter concentration (GMC) were made using automated segmentation algorithms. GMC was analyzed both voxel-by-voxel (preserving spatial precision) as well as using predefined regions of interest. Voxel-based morphometry revealed intense GMC reduction in the anterior portion relative to posterior thalami. Furthermore, thalamic atrophy was greater ipsilateral to the MTLE origin than on the contralateral side. Here we demonstrate that the thalamic atrophy is most intense in the thalamic nuclei that have strong connections with the limbic hippocampus. This finding suggests that thalamic atrophy reflects this region's anatomical and functional association with the limbic system rather than a general vulnerability to damage.

Introduction

Drug-refractory medial temporal lobe epilepsy (MTLE) is a prevalent condition with profound consequences. While visual inspection of MRI scans can typically reveal clear atrophy within hippocampal regions, less is known about damage to other regions of the brain. Here we investigate atrophy within the nuclei of the thalamus, testing the prediction that damage to different regions of the thalamus correlates with the nuclei's connections to the hippocampus.

Hippocampal atrophy is a well-known consequence of MTLE, with clear clinical relevance. Indeed, hippocampal sclerosis (HS) is the histological abnormality most frequently associated to medial temporal lobe epilepsy (Babb and Brown, 1987). Magnetic resonance imaging (MRI) permits the in vivo diagnosis of MTLE by showing signs that are associated to HS, among which the visually defined or quantitatively calculated hippocampal atrophy (HA) (Cendes et al., 1993a). By demonstrating a clear correlation with neuronal loss (Lencz et al., 1992, Spencer, 2002), the quantification of the hippocampal volume is a well-established method to investigate and quantify hippocampal damage in patients with MTLE (Andermann, 2003, Bonilha et al., 2003, Cendes et al., 1993a, Cendes et al., 1993b, Spencer, 2002).

However, the brain atrophy in patients with MTLE is not confined solely to the hippocampus. Indeed, it extends to other brain regions, such as the parahippocampal region (Falconer et al., 1964, Meencke, 1991). The damage to regions outside the hippocampus has been demonstrated by MRI studies using manual quantification of the medial temporal structures (Bonilha et al., 2004) revealing that the structures surrounding the hippocampus, such as the entorhinal and perirhinal cortex, exhibit significant atrophy (Bernasconi et al., 2000, Bernasconi et al., 2003, Bonilha et al., 2003, Salmenpera et al., 2000). Interestingly, the atrophy observed in the medial temporal lobes is more intense in structures with close anatomical and functional connections to the hippocampus (Andermann, 2003, Bernasconi et al., 2003, Bonilha et al., 2003), but can also occur in patients with MTLE who do not exhibit detectable levels of hippocampal atrophy (Bernasconi et al., 2001). Manually defined morphometry has also shown that the MTLE can result in atrophy outside of the temporal lobe. In particular, both the thalamus and the caudate nucleus exhibit a significant volume reduction in patients with MTLE (Dreifuss et al., 2001, Natsume et al., 2003).

Voxel-wise statistical comparison using automated algorithms to segment and identify gray matter has confirmed the findings of conventional manual morphometry. Voxel-based morphometry (VBM) studies performed with patients with refractory MTLE demonstrated that there is significant difference in the gray matter concentration (GMC) in patients with MTLE that is not restricted to the hippocampus, but extends to the medial portion of the temporal lobe, to the thalami and the caudate nuclei, to the cerebellum, and to other cortical areas such as the cingulated gyrus, the parietal lobes, and the insulae (Keller et al., 2002a, Keller et al., 2002b). Combining the findings from the conventional morphometry and VBM, all the medial temporal lobe regions that show reduction in volume and GMC differences are functionally or anatomically connected to the hippocampus. This corroborates the hypothesis that the damage in the brain of patients with MTLE follows a route according to a neural network of hippocampal and limbic connections, which is responsible for generating and maintaining the seizures observed in these patients (Spencer, 2002).

However, it is currently unclear whether the relationship observed within the medial temporal lobe also applies to more distant structures. For example, consider the thalamus, which shows an overall reduction of volume in patients with MTLE (Dreifuss et al., 2001, Natsume et al., 2003). Critically, the thalamus is composed of different nuclei that are connected to different regions of the brain. While the anterior nuclear group and the laterodorsal nucleus are reciprocally connected to the hippocampus and the cingulated gyrus, the other nuclear groups are not heavily connected to the limbic system (Afif and Bergman, 1997).

If the degree of neuronal damage in patients with MTLE is graded based on the level of connectivity to the limbic system, then the anterior portions of the thalamus should be more intensely affected. This finding would support the hypothesis that these regions are part of a large neuronal network involved in the generation and maintenance of seizures and therefore more prone to the neurotoxic effects of the epileptic electric discharge (Spencer, 2002, Wennberg et al., 2002). Conversely, observing equivalent atrophy in different regions of the thalamus would lend support to the notion that this region is simply more susceptible to damage than other brain areas. Here we directly test this prediction.

However, it should be noted that measuring the size of the thalamus is notoriously difficult using standard clinical quality MRI scans. Unlike the clearly defined boundaries between the medial temporal lobe cortices, the boundaries between different nuclei within the thalamus are relatively difficult to identify. Therefore, conventional manual morphometry has not been able to evaluate the neuronal damage associated to MTLE within different subregions of thalamus and has not been able to show whether the neuronal damage within the thalamus correlates with its connections to the rest of the limbic system.

Automated techniques such as VBM allow brain regions to be measured without relying on the manual delineation of the underlying structures (Ashburner and Friston, 2000). Traditional VBM statistics typically examine each and every voxel in the brain for evidence of gray/white matter atrophy. However, we can also employ regions of interest (ROIs) (Brett et al., 2002) to specifically test whether there is atrophy in preselected regions of the brain. There are two reasons why ROI analysis can offer improved statistical power compared to voxel-by-voxel analysis. First, traditional voxel-by-voxel analysis must contend with the chance of family-wise error (i.e., when making a large number of statistical comparisons, we need to control for the risk of making many false alarms). Furthermore, an ROI pools data across a large number of voxels, offering a more robust measure of the gray/white matter concentration than observed for a single voxel.

We have performed both a conventional voxel-by-voxel as well as an ROI-based VBM analysis of patients with drug-refractory MTLE. Our aim was to examine if the neuronal damage within the thalamus of patients with MTLE occurs independently of the anatomical connections, or if atrophy is more intense in the anterior regions of the thalamus that are connected to the hippocampus and the limbic system.

Section snippets

Methods

We studied 49 neurologically healthy control subjects (17 men) with a mean age of 34 years (SD = 11, range 19–60 years). We studied 43 patients with chronic refractory MTLE. Twenty-two patients had left MTLE (8 men) with mean age of 38 years (SD = 8), ranging from 18 to 54 years; and 21 patients had right MTLE (7 men) with mean age of 32 years (SD = 8), ranging from 17 to 55 years. Controls and patients had no significant difference of age [F(2,89) = 1.6, P > 0.05] or gender (Pearson's

Results

We performed three different types of voxel-based morphometry to interpret our data set. First, we performed a conventional voxel-by-voxel analysis. This analysis allows us to preserve the spatial resolution of our data. Second, we applied regions of interest to compute overall gray matter concentrations for the four predefined regions (left anterior, left posterior, right anterior, and right posterior thalamic regions). While ROI analysis lacks the spatial precision of voxel-by-voxel analysis,

Discussion

Patients with refractory MTLE exhibit gray matter atrophy affecting the thalamus. Even though there is a general reduction of the gray matter volume throughout the thalamus in patients with refractory MTLE, the reduction of GMC is more intense in the anterior–superior portion of the thalamus, as compared to posterior regions. The pattern of GMC reduction observed in the thalamus of patients with refractory MTLE supports the hypothesis that the neuronal damage observed in the brain of patients

Acknowledgments

Study supported by FAPESP and CNPq. The confection of this manuscript was inspired by an insightful comment from Prof. Robert Fisher while discussing a previous work from our group during a conference.

References (29)

  • E.H. Bertram

    Functional anatomy of spontaneous seizures in a rat model of limbic epilepsy

    Epilepsia

    (1997)
  • L. Bonilha et al.

    Medial temporal lobe atrophy in patients with refractory temporal lobe epilepsy

    J. Neurol., Neurosurg. Psychiatry

    (2003)
  • L. Bonilha et al.

    Protocol for volumetric segmentation of medial temporal structures using high-resolution 3-D magnetic resonance imaging

    Hum. Brain Mapp.

    (2004)
  • M. Brett et al.

    Region of interest analysis using an SPM toolbox

  • Cited by (109)

    • Reduction of ipsilateral thalamic volume in temporal lobe epilepsy with hippocampal sclerosis

      2018, Journal of Clinical Neuroscience
      Citation Excerpt :

      These observations support the potential role of the thalamus in TLE with HS. However, it is very controversial whether there is a progression of thalamic atrophy in TLE with HS, and so far there are few imaging studies on this issue [16–18]. In the present study, we demonstrated that the volumes of the ipsilateral thalamus in TLE patients with HS had significant negative correlation with the duration of epilepsy, although atrophy of the ipsilateral thalamus was as yet not obvious.

    View all citing articles on Scopus
    View full text