Introduction

About a quarter of patients with epilepsy fail to respond to medical treatment, experiencing an increase in morbidity and mortality compared to the general population [1]. One of the most common forms of epilepsy associated with a poorer response to antiepileptic drugs is temporal lobe epilepsy with mesial sclerosis (MTS).

In most cases of MTS, surgical treatment reduces, either in whole or in part, the frequency of epileptic seizures, improving the quality of life of drug-resistant patients. In addition, resective surgery plus antiepileptic drugs results in a lower probability of seizures than antiepileptic treatment alone [2,3,4].

Hippocampal resection is the most frequent surgical procedure in MTS. However, it requires an exhaustive presurgical evaluation to select patients who could benefit from surgery; reduce the morbidity and mortality associated with the procedure, although this is below 1% [5]; and identify the epileptogenic focus of seizures. Different non-invasive techniques have been developed to help with this assessment, providing structural and functional information about the epileptogenic focus. One such technique is magnetic resonance spectroscopy (MRS).

Magnetic resonance spectroscopy

Magnetic resonance spectroscopy is a non-invasive technique able to evaluate the products of metabolism in a living tissue, detecting possible dysfunctions and supplementing the information provided by MR. To do this, it is necessary to choose a spectrum and the region to study. The different spectra in MRS include.

Proton magnetic resonance spectroscopy (1H-MRS)

This is the most used MRS modality due to the abundance of hydrogen in the tissues of the human body. The metabolites studied in 1H-MRS include [6,7,8,9] (1) N-acetylaspartate (NAA), considered to be a marker of neuronal function as it is a product of the oxidative metabolism of neuronal mitochondria, and NAA has an osmoregulatory function and is absent in glial cells; (2) creatine (Cr) and choline (Cho), which are present in neurons and, in higher concentrations, in astrocytes, and Cr and Cho are related to intracellular bioenergetic functions; (3) Myo-inositol (mI), associated with osmoregulation and intracellular signaling; and (4) Glutamate+Glutamine (Glx), the most important excitatory neurotransmitters in the neuronal synapse. Furthermore, NAA/(Cho+Cr), NAA/Cr, NAA/Cho, and Cr/NAA are used in the evaluation of bioenergetic and neuronal functions.

Phosphorus magnetic resonance spectroscopy (31 P-RM-e) [7, 10]

This allows the measurement of phosphorus metabolites related to the energy state and membrane composition of human brain cells. This includes products of membrane synthesis such as phosphodiesters (PDE) or phosphomonoesters (PME). ATP (α, β, γ), phosphocreatine (PCr), and inorganic phosphate reflect the energy demand of the brain region studied.

Other spectra that can be used in MRS are 13C, 7Li, and 19F. However, they are only available in a few specialized research centers.

Single voxel and multiple voxel magnetic resonance spectroscopy

Once we have chosen the spectrum we then select a brain region to study [7, 9].

Single voxel magnetic resonance spectroscopy

This consists of selecting a unique voxel, which can acquire multiple shapes. The voxel must be placed in regions, which are not affected by air or water interphases. Single voxel is especially indicated in brain diseases in which imaging abnormalities are shown on MR, such as MTS.

Multiple voxel magnetic resonance spectroscopy

This allows spectroscopic data from several voxels to be obtained at the same time. It is commonly used in diseases without MR abnormalities, such as temporal epilepsy with no signs of mesial sclerosis (n-MTS)

The purpose of our systematic review was to collect the main findings obtained by MRS in MTS patients and to determine its usefulness in presurgical evaluation.

Materials and methods

A systematic literature search for MRS in MTS was made in the databases PubMed, SCOPUS, and The Cochrane Library Plus, based on the MESH terms ““Magnetic Resonance Spectroscopy”, “Proton Magnetic Resonance Spectroscopy”, “Carbon-13 Magnetic Resonance Spectroscopy”, “1H-MRS”, “31P-MRS”, “mesial temporal sclerosis”, “hippocampal sclerosis”, “mesial temporal seizure”, “mesial temporal epilepsy”, and the Boolean operators (AND, OR). We did not use additional filters. After removing 23 duplicates, there were 134 articles. We chose articles according to the PRISMA scale [11] (Preferred Reporting Items for Systematic Reviews and Meta-analyses), which is composed of phases summarized in Fig. 1.

Fig. 1
figure 1

Search strategy flowchart

Full size image

First, we performed a critical reading of titles and abstracts. Our inclusion criteria were articles in which patients with MTS were evaluated with MRS in its different modalities. In addition, we included publications in which MRS was correlated with other tests (Video-EEG, neuropsychological tests, volumetry, or histopathology).

Our exclusion criteria were (1) studies of other epilepsies, (2) studies which only included temporal epilepsy without sclerosis (n-MTS), (3) articles without an English version, (4) reviews, (5) animal studies, (6) case series, (7) studies with fewer than 10 patients with MTS, and (8) studies involving children.

After screening by title and abstract, we read the full text of the remaining 42 articles. Five articles were rejected because it was not specified that the temporal epilepsy was due to sclerosis. Three articles were excluded because there were fewer than ten cases of MTS. Two articles in which no MRS was performed and two articles in which spectroscopic evaluation was made on resected brain tissue were also excluded. Thus, 30 studies were finally included in our systematic review.

Results and discussion

We undertook a systematic review of articles reporting the spectroscopic data obtained from patients with MTS in order to examine the application of MRS in presurgical evaluation. The review finally included 30 studies, summarized in Table 1, which specifies author and year of publication, objectives, design, brain region studied, MRS characteristics, additional tests, and results. Of these 30, 23 included MRS as part of the presurgical evaluation of patients with MTS [10, 12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33], while 2 studies performed MRS after surgery [34, 35]. The remaining 5 did not specify whether MRS was included for presurgical or postsurgical assessment [36,37,38,39,40].

Table 1 Studies with Magnetic Resonance Spectroscopy in Mesial Temporal Sclerosis
Full size table

MRS studies in ipsilateral hippocampus

The main findings were found in the ipsilateral hippocampus, showing a decrease in NAA (13/30 studies) [12, 13, 15, 19, 21, 23,24,25, 28, 29, 31, 32, 38], a decrease in the ratio NAA/Cho+Cr (9/30 studies) [17, 18, 30, 33, 34, 36, 37, 39, 40], and a decrease in the ratio NAA/Cr (8/30 studies) [14, 16, 20, 22, 26, 27, 35, 39]. MTS was traditionally associated with a neuron loss in the hippocampus [23]. More recent studies hypothesized that there is a relationship between MTS and neuronal mitochondrial dysfunctions [41]. NAA is a product of the Krebs Cycle, which takes place in the mitochondria, so the alterations observed in these 29 articles seem to be consistent with the mitochondrial dysfunction theory. In addition, as NAA has a role in the osmoregulation of neurons [8], low NAA levels cannot maintain the integrity of these cells, resulting in neuron loss.

The decrease in NAA can be detected in different ways, for example, measuring just NAA or determining its different ratios. The ratios that are most altered are NAA/Cho+Cr and NAA/Cr. However, when a decrease is detected in NAA/Cho+Cr or NAA/Cr, it is questioned whether the ratio is reduced due to theoretical high levels of Cr or Cho (which are in the denominator of the ratio, so an increase in their levels could result in a decrease of the ratio) instead of altered NAA levels. Accordingly, it is necessary to obtain the individual results of Cr and Cho, as did the studies summarized in Table 1.

In most studies, patients with MTS did not show a statistically significant alteration in Cr or Cho levels [12, 13, 16,17,18,19,20,21,22, 27, 30,31,32,33,34,35,36,37, 40]. The only exception was the study by Riederer et al. [15], in which Cr was decreased, contrary to what was expected. However, this study admitted a few limitations in spectroscopic data acquisition due to artifacts of bone and air from areas adjacent to the selected voxel.

Comparative studies in extrahippocampal regions

MTS may affect other brain regions outside the hippocampus, as seen in several of the studies included in our systematic review. A decrease in NAA levels was detected in frontal and occipital lobes [36], thalamus [39], temporal operculum, cerebellum, insula [21], and temporal lobe white matter of the ipsilateral side [13, 27]. In addition, Park et al. [10] noted that additional mitochondrial products such as inorganic phosphate and the PCr/γ-ATP were decreased in the insula and basal ganglia using 31-MRS. In 9 studies, NAA was decreased on the contralateral side, and the most altered region on this side was the hippocampus [12, 15, 18, 20, 31, 32, 34, 35, 39].

Studies in which MRS was performed for surgical evaluation

Eleven studies found a decrease in NAA, NAA/Cho+Cr, and NAA/Cr ratios in presurgical evaluation [12, 14,15,16, 18,19,20,21,22, 25, 40].

MRS was performed as postsurgical evaluation in 2 studies, which concluded that the damage produced by MTS could be caused by functional and dynamic mechanisms. Spencer et al. [35] only included postsurgical MRS, while Lantz et al. [34] analyzed spectroscopic data obtained both before and after surgery.

Lantz et al. [34] noted that patients with MTS who responded to surgical treatment showed MRS normalization of the contralateral abnormalities detected in the preoperative evaluation, with NAA values almost reaching those found in healthy subjects. This suggests that the decrease in NAA could be reversible and modified by such treatment as hippocampal resection. Also, postoperative normalization was more pronounced for patients showing an extensive decrease in NAA [34]. The explanation for this is still unknown. However, postsurgical normalization in MRS could be more appreciable in patients who have instability in their clinical condition and their metabolic state, so they have more room for improvement as compared to patients with disease restricted mainly to the hippocampus. If this association is correct, presurgical MRS could allow the selection of patients who could benefit more from hippocampal resection.

Comparative MRS studies with clinical characteristics and epileptiform activity on EEG

Concerning the possible association between MRS and seizures, 6 studies showed that lower NAA correlated with an early onset age of the symptoms [32], longer duration of epilepsy [20, 32, 37], greater frequency of seizures [30], and epileptogenic activity on EEG on the ipsilateral and contralateral sides [20].

Focusing on studies which evaluated the response to surgery, Spencer et al. [35] noted that patients with a higher frequency of seizures after surgery presented more brain regions with low NAA in postsurgical MRS. Therefore, postsurgical MRS could predict the long-term response in patients with MTS, although comparative studies with MRS evaluation before and after surgery are necessary.

MRS studies that included neuropsychological evaluation

One of the most common and disabling symptoms in MTS is dysfunction in verbal memory, more severe in patients with left MTS. Four studies detected a decrease in NAA [24, 28, 29] and a decrease in NAA/Cho+Cr [37] in patients with a worse performance in verbal memory tests.

Regarding patients with right MTS, no cognitive deficiencies related to verbal memory could be detected, except in the study of Mantoan et al. [37], in which patients with right MTS who had low NAA/Cho+Cr levels in the left hippocampus had worse scores in verbal memory.

Finally, Martin et al. [29] included facial recognition tests, with the worst scores detected in patients with a decrease in NAA in the right hippocampus.

Comparative studies of patients with and without sclerosis

Several comparative studies included in this review noted that, as well as MTS, n-MTS patients also presented a decrease in NAA [12, 19, 38] and a decrease in NAA/Cho+Cr [40], in the ipsilateral hippocampus, although NAA was even lower in the MTS patients.

According to some studies, to distinguish between MTS and n-MTS, we should focus on other metabolites. The most striking difference was in the level of Glx, which was significantly increased in 2 studies. Woermann et al. [31] found this in the ipsilateral hippocampus and Simister et al [22] detected it on the contralateral side. However, Doelken et al. [38] and Riederer et al. [15] did not find any significant alteration in Glx in n-MTS patients. Though more studies including this metabolite are necessary, Glx could possibly differentiate between MTS and n-MTS.

Regarding mI, the results from 6 studies were conflicting; 4 found no statistically significant differences between MTS and n-MTS [12, 13, 22, 38]. On the other hand, Riederer et al. [15] noted that n-MTS patients presented a lower mI/Cr ratio than MTS patients, while Mueller et al. [19] only detected low mI levels in MTS patients.

Single voxel and multiple voxel MRS

Single voxel MRS was performed in 18 studies [12,13,14,15,16,17,18, 20, 25, 27, 28, 30,31,32,33, 37,38,39], proving technically easier to analyze the hippocampus. Patients with MTS were evaluated using multiple voxel in the remaining 12 studies [10, 19, 21,22,23,24, 26, 29, 34,35,36, 40], mainly used to study extrahippocampal regions in which it could be difficult to place the voxel due to the absence of structural abnormalities on MR [9]. However, none of these studies compared single and multiple voxel MRS in patients with MTS.

Limitations and future lines

The main limitations of the studies analyzed in this review were the heterogeneity in the MRS technique, the sample sizes, and the characteristics of the population. We included only two studies performed MRS after surgery, so it is not possible to conclude the usefulness of MRS in post-surgical evaluation. Accordingly, this field requires further studies that include presurgical and postsurgical MRS evaluation in a larger sample size to understand the main reasons for the postoperative normalization in the contralateral hippocampus. In addition, it is necessary to establish a protocol for MRS evaluation, using the same ratio of NAA in order to accurately compare between different studies. Future studies all need to include quantitative and qualitative variables such as the frequency of seizures, onset age, duration of epilepsy, and epileptogenic activity on EEG. Only then will we be able to determine the possible association between these variables and MRS. In addition, measurement of other metabolites such as Glx and mI is also required in the MRS evaluation.

Conclusions

This systematic review shows that, in most MRS studies, patients with MTS presented a decrease in NAA in the ipsilateral hippocampus. Furthermore, the decrease in NAA could be detected in extrahippocampal regions on the ipsilateral and the contralateral sides. Additionally, comparative studies also presented NAA decreased in the ipsilateral hippocampus in n-MTS patients. Thus, MRS could be useful but not specific for the diagnosis of MTS and could have a role in the presurgical evaluation to locate the epileptogenic focus. However, further research in this field is needed.