PERSPECTIVAS DEL ESTUDIO

5.7.1 Temporal lobe epilepsy 5.7.2 Extratemporal epilepsy

5.8 Discussion

5.1 Introduction

In magnetic resonance, the resonance frequency of any nucleus is proportional to the local magnetic field that it experiences. The local field is influenced by the electrons within the vicinity o f a nucleus, and therefore the resonance frequencies o f individual nuclei depend on their chemical environment. The relative shift of resonance frequencies in different chemical environments is termed chemical shift and is expressed as dimensionless units o f parts per million. Magnetic resonance spectroscopy (MRS) enables the detection of signals from nuclei within different compounds, and therefore provides a means of detecting chemical abnormalities in tissue, and 'H are the nuclei that have been most

widely used in clinical studies in view of their high natural abundance and relatively high sensitivity.

Phosphorus magnetic resonance spectroscopy MRS) was originally o f particular interest in view of its ability to monitor energy metabolism, by detection of signals from ATP, phosphocreatine (PCr) and inorganic phosphate (Pi). Studies initially concentrated on muscle, but later it was evident that useful cerebral changes could be seen in hypoxic ischaemic encephalopathy in neonates (Cady et a l, 1983; Hope et a l, 1984; Younkin et a l , 1994). A fall in energy status can be demonstrated in the brain of affected infants, seen predominantly as a fall in the PCr/Pi ratio. This has been correlated with neuro- developmental outcome (Roth et a l, 1992). ^*P MRS studies have also been performed in a small number of adults with temporal (Hugg et a l, 1992; Kuzniecky et a l, 1992) and frontal lobe (Garcia et a l, 1994; Laxer et a l, 1992) epilepsy, but results have been inconsistent. An increased inorganic phosphate has been demonstrated on the side of the seizure focus in TLE (Hugg et a l, 1992; Kuzniecky et a l, 1992), but no such change has been found in frontal lobe epilepsy (Garcia et a l, 1994). Furthermore, one group has demonstrated an increase in pH and decreased phosphomonoester ipsilateral to the seizure focus in temporal and frontal lobe epilepsy (Garcia et a l, 1994; Hugg et a l, 1992) but this was not confirmed in TLE by a second group (Kuzniecky et a l, 1992).

Although it initially posed more technical difficulties than ^’P MRS, proton magnetic resonance spectroscopy (*H MRS) is now used extensively for the study of brain tissue.

concentrations than water, and for this reason the water, and (to a much lesser degree in the brain) fat, need to be suppressed in order to be able to detect the metabolites of signals.

An example o f a MR spectrum from the brain is shown in Fig. 5.1. The dominant contributions to the spectrum are from N-acetylaspartate (NAA) at 2.0ppm, creatine + phosphocreatine (Cr) at 3.0ppm and choline-containing compounds (Cho), phosphoryl- and glycerophosphorylcholine seen at 3.2ppm. NAA has been shown from lesional (Roller et a l , 1984), tumour (Sutton et a l, 1992) and cell culture (Urenjak et a l, 1992; Urenjak et a l, 1993) studies to be located primarily in neurons, and although also found in high quantities in oligodendrocyte type II astrocyte progenitor cells, it is thought to be a marker o f neuronal integrity (Urenjak et a l, 1992; Urenjak et a l, 1993). Cr and Cho are found both in neurons and astrocytes, although cell studies suggest that they are present in higher concentrations in astrocytes (Urenjak et a l, 1993).

Proton MRS has been investigated in adults with temporal and frontal lobe epilepsy (Cendes et a l, 1994; Connelly et a l, 1994; Cook et a l, 1991; Gadian et a l, 1994; Garcia

et a l, 1995; Hugg et a l, 1993; Ng et a l, 1994; Ende et al., 1997). Frontal lobe studies have been performed on a small number of patients; single voxel techniques detected no significant abnormalities (Cook et a l, 1991) whereas chemical shift imaging (CSI), recording data from multiple voxels within a slab of tissue in eight patients, demonstrated a reduction in the NAA/Cr ratio of between 5-30% ipsilateral to the seizure focus (Garcia

B

NAA

Cho

^ .0 3 .5 3.0 2.5 2.0 1.5 1.0 0 .5

Figure 5.1: Coronal magnetic resonance image (A) and a H spectrum (B) from a normal subject. The boxes indicate the position of the 8ml cubic volume of interest from which spectra were obtained. The dominant

contributions to the spectrum are from N-acetylaspartate (NAA), creatine plus phosphocreatine (Cr), and choline containing compounds

In adults with TLE, a reduction in NAA ipsilateral to the seizure focus, suggestive of neuronal loss or dysfunction, has been shown to aid in latéralisation. In addition, some studies have shown a high frequency of bitemporal abnormalities (Connelly et al., 1994; Ng et a l, 1994). The single voxel technique used for the studies described in this chapter uses 2x2x2cm regions that include a substantial portion of the mesial temporal regions with only a minor contribution from the hippocampus (Fig 5.1). It may be considered therefore to complement focal information obtained about the hippocampus from T2 relaxometry. The purpose of the part of this study described in this chapter was to determine the incidence o f abnormalities as detected by single voxel ’H MRS o f the mesial temporal lobes in children with nonlesional TLE (as defined in Section 4.5.1), to determine whether a similar degree o f abnormality was found to that reported in adults, and to investigate the role of 'H MRS data in latéralisation. The study was extended to address the question of whether abnormalities of 'H MRS of the mesial temporal regions were seen in extratemporal epilepsy.

5.2 Methods

Spectra were obtained from 2x2x2cm cubes centred on the medial portions of the right and left temporal lobes, as shown in Fig. 5.1. Spatial localisation was achieved using a 90°-180°-180° spin echo technique, with three selective radiofrequency pulses applied in the presence o f orthogonal gradients o f 2mT/m. Water suppression was achieved by pre­ irradiation o f the water resonance using a 90° Gaussian pulse with a 60Hz bandwidth, followed by a spoiler gradient. TR was 1600ms and TE 135ms. After global and local shimming, and optimization of the water suppression pulse, data were collected in 2-4 blocks o f 128 scans. The time domain data were corrected for eddy-current induced phase

modulation using non-water-suppressed data as a reference (Klose, 1990). Exponential multiplication corresponding to 1-Hz line broadening was carried out prior to Fourier transformation, and a cubic spline baseline correction was performed.

Signal intensities for each of the relevant compounds demonstrated in Fig. 5.1 were measured from the peak areas by integration. The effects o f XI and T2 relaxation were not determined, and therefore it was not possible to convert these intensities to concentra­ tion measurements. However, by multiplying the observed signal intensities by the 90° pulse voltage, it was possible to compensate for differences in radiofrequency coil loading (Austin et a l, 1991; Hoult and Richards, 1976) and thereby to compare absolute signal intensities between different subjects. The intensity ratio is dimensionless by definition, while the absolute intensities are given in arbitrary units that reflect the settings of the magnetic resonance system used. Data are presented in the form of these corrected signal intensities, and also in terms of the intensity ratio NAA/(Cho+Cr). Choice of this

particular ratio is based partly on the observation that the group data show a significant mean decrease in NAA and increases in the Cr and Cho signals. In addition, in the temporal lobe it is difficult in some cases to achieve full spectral separation of the Cr and Cho signals due to particular problems with magnetic field inhomogeneity in this region. The NAA/(Cho+Cr) ratio is unaffected by this latter problem.