Group comparisons

Variation was observed within both the control and TLE patient MTL and also across these groups in the anterior MTL voxel. Several metabolites co-varied with grey matter content.

In the control group Cho concentrations were higher in the anterior MTL. Cr was non- significantly lower in the middle voxel and NAAt did not vary significantly. This resulted in lower NAAt/(Cr+Cho) anteriorly and higher NAAt/Cr in the middle voxel. GLX and Ins were higher in the anterior voxel. These effects were reduced by the incorporation of grey matter proportion as a co-variate, such that NAAt/(Cr+Cho) and GLX no longer varied significantly. Other workers have considered metabolite variation in the control temporal lobe. Vermathen et al(Vermathen et al., 2000) used MRSI at TE=135ms, and found lower NAA/(Cr+Cho) and higher Cho in the anterior control MTL although no correction for grey matter proportion was made. Another group has described lower NAA/Cr in the hippocampus than the lateral temporal lobe(Wang et al., 2008). Lower NAAt/Cr was also seen in the posterior MTL than in the posterior LTL in all the groups, due to a lower Cr in the lateral voxel, without significant variation in NAAt or Cho. This is consistent with a study of metabolite concentration in grey and white matter in voxels outside the temporal lobe, at the same short echo time (TE=30ms)(McLean et al., 2000). In this study the ratio of Cr concentrations in grey versus white matter was 1.7, compared with 1.16 for NAAt and

0.9 for Cho. This implies that Cr concentration will be most altered by a reduction in the proportion of grey matter. The concentration of Ins and GLX were lower in the posterior LTL than the posterior MTL in the control group. This would also be in keeping with the observed metabolite ratios for GLX and Ins between grey and white matter of 2.4 and 1.6 respectively(McLean et al., 2000) implying that there would be lower concentrations in regions with lower proportions of grey matter.

In the HS group the anterior ipsilateral voxel showed reduced NAAt, but also low Cr and Cho (Figure 3.1.2). Other studies have similarly reported reduction in both NAA and Cr in the sclerotic hippocampus(Ende et al., 1997;Duc et al., 1998) and a histological examination in an animal model of temporal lobe epilepsy(Tokumitsu et al., 1997) identified reduced NAA and Cr in the presence of neuronal loss and gliosis. The single voxel study(Woermann et al., 1999c) did not identify significant group effects for Cr or Cho but averaged the entire hippocampus, which could have masked changes confined to the anterior region. It has been held that in the sclerotic anterior hippocampus there is loss of neuronal tissue with consequent loss of NAAt, because of the localisation of NAA in neurones(Urenjak et al., 1993), and reactive gliosis, with consequent increase in Cr and Cho. This understanding was largely based on single voxel studies or MRSI at longer TE. The finding that all three metabolites were decreased in this region might indicate that Cr and Cho are sufficiently present in neurones to also be affected by the presumed process of neuronal loss/dysfunction(Urenjak et al., 1993).

In agreement with the above trends for NAAt, Cr and Cho the ratios NAAt/Cr and NAAt/(Cr+Cho) were unhelpful in identifying abnormality in the TLE groups. This concurs with the findings of the single voxel study(Woermann et al., 1999c). Other groups, however, and usually with longer TE, have found these ratios to be sensitive(Cendes et al., 1994;Vermathen et al., 1997;Achten et al., 1997). The reason for this difference in findings is not established but these studies were in general performed with different acquisition parameters and utilise different descriptions of abnormality. At short TE there is improved signal to noise which might make quantification more reliable, and although the baseline at short TE is more variable

there are established spectral peak fitting programmes that produce reliable and repeatable estimates of metabolite concentration. It is also possible that metabolite T2 might be altered in pathological tissue (e.g. hippocampal sclerosis), in which event those studies performed at longer TE would be more likely to introduce metabolite measurement error.

The single voxel study identified elevated GLX in the mesial temporal lobe ipsilateral to seizure onset in the MRIN group. In the current study ipsilateral elevation was not found but elevation of GLX in the anterior voxel of the MRIN contralateral group was observed. It is of interest that both studies identified elevated GLX in patients with TLE and normal MR imaging. However several factors limit the interpretation of these findings. Firstly GLX represents a signal complex of glutamate plus glutamine and it is not currently possible to determine whether elevation in GLX represents change in glutamate, in glutamine or in both metabolites. Also GLX is measured less reliably than NAAt and in these two studies the number of subjects is relatively small. Furthermore the two studies measured GLX in different volumes (total hippocampal GLX versus anterior MTL GLX in the current study) with a very limited number of subjects participating in both studies.

In the MRIN ipsilateral group NAAt was significantly lower in the posterior LTL than in the posterior MTL. This finding might indicate involvement of lateral neo-cortex in a pathological process in this group and agrees with a previous MRS study of similarly selected subjects which identified extensive metabolite abnormality in the temporal lobe ipsilateral to seizure onset compared to healthy controls(Connelly et al., 1998).

In this study no relationship between NAAt and seizure frequency, total lifetime secondary generalised seizures, or time since last seizure was noted. This is consistent with the single voxel findings(Woermann et al., 1999c) and also with a longitudinal study, which did not observe change in the ratio NAA/Cr in patients who had become seizure free on anti-epileptic drug therapy(Li et al., 2000c). Another study(Maton et

al., 2001b) identified no consistent changes in metabolite peak ratios between the inter-ictal and post-ictal state in TLE.

MRSI has a demonstrated role in the study of subjects with epilepsy. The ideal study would provide a full metabolite profile as well as spatial information. This current short TE work is the first MRSI study to quantify Ins and GLX as well as NAA, Cho and Cr at different positions along the MTL and the lateral temporal lobe, in the normal temporal lobe and also in subjects with temporal lobe epilepsy. The study identifies how the obtained concentrations are influenced separately by variation in voxel position and tissue composition and show how such variability might limit the interpretation of metabolite ratios.