Correlation of magnetic resonance spectroscopy and magnetic resonance imaging with findings of electroencephalography in patients with temporal lobe epilepsy

Abstract

Introduction

This research aimed to see how well magnetic resonance spectroscopy (MRS) could identify the lateralization side in individuals with temporal lobe epilepsy (TLE) compared to electroencephalography (EEG) and magnetic resonance imaging (MRI) results.

Methods

Twenty‐three individuals were included in this research and diagnosed with TLE (both clinically and by EEG). Clinical exams, interictal EEG, and MRI were performed on all patients. In addition, the individuals were also subjected to proton MRS.

Results

The age range of 23 participants was 20–55 years (mean = 34.6 ± 8.5); 10 were male (44%), and 13 were female (56%). The right temporal lobe MRI showed a sensitivity and specificity of 60% and 55% for detecting mesial temporal lobe sclerosis (MTS) foci, respectively (positive predictive value (PPV) of 27% and negative predictive value (NPV) of 83%). MRI showed 83% sensitivity and 35% specificity for MTS foci in the left temporal lobe (PPV of 31% and NPV of 86%). MRS showed 61% sensitivity and 100% specificity in the right temporal lobe (PPV 100%) and 80% sensitivity and specificity in the left temporal lobe (PPV 100%) for identifying MTS foci. Overall, MRS (both left and right) results matched EEG findings.

Conclusion

MRS is a potential noninvasive neuroradiology technique for assessing epilepsy patients because it is more sensitive than structural MRI in identifying MTS. The results of the study overall appears to be of interest but still need further support from future studies with larger sample sizes.

Magnetic resonance spectroscopy is a potential noninvasive neuroradiology technique for assessing epilepsy patients because it is more sensitive than structural magnetic resonance imaging in identifying mesial temporal lobe sclerosis.

Introduction

Epilepsy is more common in developing and tropical countries, with prevalence rates ranging from 14 to 57 cases per 1000 people. ¹ The most common cause of focal and refractory seizures is temporal lobe epilepsy (TLE). ² Temporal lobe epilepsy with seizures originating from the mesial temporal lobe structure is referred to as mesial temporal lobe epilepsy (MTLE), and it accounts for the majority of instances of drug‐resistant epilepsy. ³ The most often seen underlying pathology in individuals with temporal lobe epilepsy is mesial temporal sclerosis (MTS), characterised by neuronal loss and gliosis in the medial temporal lobe. ⁴ Patients with MTS often have epilepsy of the medial temporal lobe with focal seizures originating in the amygdalohippocampal complex. ⁵ In addition, T1‐weighted hippocampus atrophy and an enhanced T2‐weighted and fluid‐attenuated inversion recovery signal change are seen in MTS magnetic resonance imaging (MRI) findings. ⁶

It is critical to accurately localise the epileptogenic zone in order to provide surgical resection. MRI and video electroencephalographic (V‐EEG) monitoring were common diagnostic methods. MRI can be used to confirm the lesion with good accuracy (up to 90% sensitivity and 85% specificity). However, in 29% of partial epileptic patients, MRI has been found to be non‐contributive. ⁷ This is the primary rationale for the increased use of additional neuroimaging modalities in epilepsy surgery candidates today. Multimodal presurgical treatments have been shown to be very beneficial in localising and lateralizing seizure foci. Positron emission tomography (PET), single‐photon emission computed tomography (SPECT), functional magnetic resonance imaging (fMRI), voxel‐based morphometry (VBM), diffusion‐weighted imaging and proton magnetic spectroscopic imaging (¹H‐MRS) are popular advanced neuroimaging techniques that have previously been used in epileptic patients. ⁸

Proton magnetic resonance spectroscopy elucidates the biochemistry of particular brain areas of interest and visualises changes in metabolite concentrations prior to the appearance of structural abnormalities. It is sensitive to identifying metabolic changes in dysfunctional epileptogenic areas, such as the hippocampus formation in people with TLE before the onset of increased T2‐signal intensity, volume loss and contrast enhancement. ⁹ In addition, proton magnetic resonance spectroscopy may improve diagnostic sensitivity in lateralization of epileptic foci by showing reduced N‐acetyl‐aspartate (NAA), increased glutamate, and glutamine (Glx) and Myo‐inositol (Myo‐inositol). ^{10 1}H‐MRS has a 50% specificity and a 96% sensitivity for properly lateralizing epileptogenic zones (hippocampal and amygdala). However, it is beneficial to identify the underlying neoplastic pathology and predict seizure control in the long‐term follow‐up period after surgery. ¹¹ To evaluate the MRS data regarding lateralization and localization of the epileptogenic zone in patients with epilepsy, several studies compared H‐MRS findings to high‐resolution MRI, PET and SPECT, ¹² demonstrating that 1H‐MRS is more sensitive than other imaging techniques such as PET. ¹³ At the same time, it correlates more strongly with EEG data as the primary tool for the clinical diagnosis of TLE. ¹¹

This research aimed to determine the ability of MRS and MRI to identify the MTS lateralization taking into account the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) in patients with TLE in relation to EEG results.

Methods

The study included 23 individuals diagnosed with temporal lobe epilepsy (clinically and electroencephalographically) while admitted to neurologic wards at Kashani Hospital from 2019 to 2020. TLE was diagnosed based on the clinical history and seizure description in accordance with the International League Against Epilepsy (ILAE) 1989, ¹⁴ which defined TLE as a type of localization‐related epilepsy with typical clinical and electroencephalographic characteristics. It can manifest as simple partial, complex partial, secondary generalised or a combination of the above.

Patients were excluded if an imaging study (MRI) revealed any focal abnormalities other than MTS. Age under 18, previous history of trauma and metabolic diseases were other exclusion criteria.

All patients and controls included in this study underwent a full clinical neurological assessment and an MRS (N‐acetyl aspartate to creatine ratio (NAA/Cr), NAA/Cr + choline (Cho) ratio, NAA) examination. In addition, patients had video scalp EEG and brain MRI.

All patients underwent electroencephalography (EEG) utilising the Nihon Kohden device. This system features a typical 64‐channel. The EEG is recorded for an extended period throughout this process, accompanied by continuous closed‐circuit video observation. Simultaneous presentation of the digitalized EEG and recorded behaviour enables point‐to‐point linkages between recorded events and any accompanying electrographic changes.

During video‐EEG monitoring, the patient wears an EEG transmitter connected to a coaxial cable. Continuous behavioural surveillance is possible with wall‐mounted video cameras. Both the EEG and video signals are transferred to the EEG recording device. Both the EEG and video signals are shown concurrently for online inspection, and both are stored on a hard drive or DVDs.

The full EEG evaluation was reviewed, and a sharp wave was classified as epileptiform if it had a sharp shape, a duration of 200 ms, and was distinguishable from the EEG background. Unitemporal interictal discharges (IEDs) necessitated at least 80% lateralization. Patients with 80% of interictal EEG onset lateralized to one side were classified as having bitemporal IEDs.

The MRI was performed on a 1.5 T clinical machine (Siemens Magnetom Avanto). We got the following image data sets: TR/TE/TI = 1630/2.82/1100 ms, 15° flip angle, 1 mm slice thickness, FOV = 256,192, TR/TE = 5200/92 ms, slice thickness 3 mm with gap = 3.3 mm, FOV = 180,180, FLAIR TR/TE/TI = 7800/82/2300.8 ms, slice thickness 3 mm, gap = 3.3 mm, FOV = 163,190.

The proton MR spectroscopic imaging investigations were performed using a 1.5 T Siemens Magnetom Avanto system with the following parameters: 3D Point Resolved Spectroscopy (PRESS) with additional phase encoding, including both hippocampi TR/TE = 1200/135 ms, flip angle 90, vector size = 512, number of phase encoding steps = 12, resolution 5.62 8.125 8.75 mm and total acquisition time = 7 min. The PRESS box was selected to provide coverage of the temporal lobes. A single voxel in white matter or grey matter (just one kind of tissue) was acquired with the following settings for correction of eddy current artefact: TR/TE = 1000/30 ms, flip angle = 90, vector size = 512 and acquisition duration = 5 minutes. FLAIR pictures in coronal view were utilised to determine the location (TR/TE/TI = 7800/82/2300.8 ms).

To be eligible for inclusion in the research, informed parental permission was acquired. The study was conducted under the guidelines of the Isfahan University of Medical Sciences, Local Ethics Committee (IR.MUI.MED.REC.1399.214).

To determine the diagnostic value and compatibility of MRS and MRI findings (both left and right) with EEG, the parameters of sensitivity, specificity and positive and negative predictive value (PPV and NPV) were used. Based on a 95% confidence interval, the diagnostic value of MRS, MRI and EEG was analysed with Medcalc version 19.5.6. SPSS (version 26) software program was used to analyse patients' demographic and clinical observations (descriptive statistics and chi‐square tests).

Results

The age range of 23 participants was 20 to 55 years (mean = 34.6 ± 8.5); 10 were male (44%), and 13 were female (56%). The demographics of the study participants are summarised in Table 1.

Table 1.

Demographic characteristics of patients.

Characteristics	Frequency (%)
Sex
Female	13 (56%)
Male	10 (44%)
Site of MTS in MRS
Lateral	15 (65%)
Right	3 (13%)
Left	5 (22%)
Site of seizure in EEG
Right	11 (48%)
Left	12 (52%)

EEG, electroencephalography; MRS, magnetic resonance spectroscopy; MTS, mesial temporal sclerosis.

Table 2 demonstrates the frequency of negative and positive MRS, MRI and EEG results in both temporal lobes (in the case of detecting the MTS). MRI returned negative results for 74% and 78% of scans for the left and right temporal lobes, respectively, and MRS returned negative results for 13% and 22% of scans for the left and right temporal lobes, respectively. MRS returned positive results for 87% and 78% of scans for the left and right temporal lobes, respectively, and MRI returned positive results for 26% and 22% of scans for the left and right temporal lobes, respectively (Table 2).

Table 2.

Frequency (%) of negative and positive results in epileptic patients.

	Left_MRI	Right_MRI	Left_MRS	Right_MRS
Negative	17 (74%)	18 (78%)	3 (13%)	5 (22%)
Positive	6 (26%)	5 (22%)	20 (87%)	18 (78%)

MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy.

Through sensitivity and specificity evaluations of MRI and MRS in relation to EEG findings, it was found that MRI in the right temporal lobe had sensitivity and specificity of 60% and 55% for finding the MTS foci (PPV of 27% and NPV of 83%). MRI had 83% and 35% sensitivity and specificity in the left temporal lobe for MTS foci (PPV of 31% and NPV of 86%). On the other hand, MRS had 61% and 100% sensitivity and specificity in the right temporal lobe (PPV of 100% and NPV of 42%) along with 80% and 100% sensitivity and specificity in the left temporal lobe (PPV of 100% and NPV of 43%) for detecting MTS foci. Overall, MRS findings (both left and right) were most consistent with EEG findings. Although MRS showed high specificity, its low sensitivity showed a poor ability to screen MTS foci, and other diagnostic tools are required to rule out a negative result (Table 3).

Table 3.

The area under the ROC curve (AUC) for the right and left temporal lobe in MRI and MRS (Gold Standard: EEG).

Variables	MRI		MRS
	Right	Left	Right	Left
Area under the ROC curve (AUC)	0.578	0.593	0.806	0.900
95% Confidence interval (CI)	0.356 to 0.778	0.371 to 0.791	0.589 to 0.939	0.703 to 0.985
Significance level P (Area = 0.5)	0.5688	0.3637	<0.0001	<0.0001
Youden index J	0.1556	0.1863	0.6111	0.8000
Associated criterion	>0	>0	>0	>0
Sensitivity	60.00	83	61	80.00
Specificity	55	35	100.00	100.00
+PV (95% CI)	27 (13.4–47.5)	31 (21.6–42.9)	100.00	100.00
−PV (95% CI)	83 (61.3–94.0)	86 (47.3–97.6)	42 (28.6–56.0)	43 (23.8–64.3)

EEG, electroencephalography; MRI, magnetic resonance imaging; MRS, magnetic resonance spectroscopy; +PV, positive predictive value; −PV, negative predictive value.

Discussion

MTS is the primary cause of refractory epilepsy; according to the final report of the epilepsy institutional review board, 55.5% of all epileptic patients were diagnosed with MTS, which corresponds to reports in the literature. ¹⁵ , ¹⁶ Exact lateralization of the seizure focus with a noninvasive study is critical in managing temporal lobe epilepsy because surgical resection of the epileptogenic lesion results in a favourable outcome. On the other hand, failure to lateralize the foci with noninvasive examination may result in an invasive study or additional surgery for the placement of intracranial electrodes, which may carry potential risks. ¹⁷ However, several causes of temporal lobe epilepsy, such as hippocampal sclerosis and developmental abnormalities, may go undetected, even when using a range of MR imaging methods, such as MR volumetry and T2 relaxation time measurement. ¹⁸ Although conventional MRI, even with high‐resolution thin sequences, has reported low predictive values with 42% sensitivity and 80% specificity for detecting hippocampal atrophy with a high rate of false negatives, ¹⁵ they tend to improve in patients with moderate to severe MTS, where values of up to 93% sensitivity and 95% specificity have been reported. ¹⁹ The current investigation found values within the reported ranges, with a sensitivity of 60% and specificity of 55% for the right temporal MTS and 83% sensitivity and 35% specificity for the left temporal lobe.

Quantitative approaches based on imaging investigations have been the focus of study for detecting hippocampal atrophy during the last several decades, resulting in its use as a prediction tool for the existence and severity of this illness, particularly in challenging situations. ²⁰ In a study on patients undergoing surgery for MTS using temporal pole MRS, they demonstrated changes in Cho, Cr and NAA concentrations, as well as NAA/Cho and NAA/Cr ratios, in the hippocampus ipsilateral to the MTS, with a sensitivity of 100%, a specificity of 80%, a PPV of 87% and an NPV of 100%. ²⁰ In another study that evaluated the MRS ability in lateralizing MTS compared with PET and MRI, they found a sensitivity of 85%, a false negative rate of 15% and a false positive rate of 3%. ²¹ Guo et al. discovered statistically significant differences in cerebral blood flow between patients with MTS and normal subjects using perfusion techniques with arterial spin labeling (ASL), which, when combined with conventional MRI, resulted in a sensitivity of 96.15% for lateralization of MTS. ²²

Based on our MRS results, we found 61% and 100% sensitivity and specificity in the right temporal lobe (PPV of 100% and NPV of 42%) along with 80% and 100% sensitivity and specificity (PPV of 100% and NPV of 43%) in the left temporal lobe for detecting MTS foci. This particular study demonstrated that MRS lateralized MTS in both lobes with a specificity of 100%. Most imaging modalities focus on temporal lobe epilepsy because surgery can improve outcomes in many patients. The only effective treatment is surgical resection of the affected hippocampus and anterior temporal lobe, which cures up to 90% of cases but requires accurate pre‐operative lateralization of the epileptic focus. Recurrent seizure attacks resistant to medication that significantly impact their life, family and society are candidates for surgery. ²³ So MRS's role in lateralizing MTS foci could bring robust mapping for surgeons before surgery for MTS. In addition, MRS is a noninvasive modality that provides metabolic information about brain tissue and biochemical tissue characterisation beyond MRI. Metabolic abnormalities often precede structural changes, so MRS can detect MRI‐invisible abnormalities. ²³

This study aimed to determine whether proton MRS can detect the epileptic focus (side of lateralization) in patients with TLE compared to EEG. We discovered that ¹HMRS detected MTS 87% in the left temporal lobe and 78% in the right temporal lobe. MRI revealed MTS of 26% in the left temporal lobe and 22% in the right temporal lobe, and EEG detected MTS of 70% in the left temporal lobe and 48% in the right temporal lobe. Similar to us in a study on the ability of ¹HMRS and MRI to localise the side of epileptic focus in comparison with EEG, it was discovered that in patients with unitemporal localised MTS in EEG (20 patients), the ¹HMRS abnormalities (decreased NAA/Cr ratio) were lateralized in 19 out of 20 patients (95%) as it was unilateral in 16 patients and bilateral with left or right predominance in 3 patients, and MRS was normal in 1 patient, while MRI revealed abnormalities in 11 (55%) of the 20 patients, the remaining nine patients had nonlateralized lesions, and five patients had normal scans. ²⁴ Also, MRI could not detect more MTS lateralization in our patients, while ¹HMRS detected with more frequency. In line with us in a study on ¹HMRS in MRI‐negative temporal lobe epilepsy, MRS detected metabolic problems in patients with normal MRI, and the researchers concluded that MRS adds value to MRI and improves the sensitivity of global MR examinations. ²⁵

This research proves that ¹HMRS had more specificity to lateralization detection. Therefore, we may explain that the MRS adds something to the diagnosis in these circumstances but not enough to prompt a surgical strategy.

Limitations

In this study, the number of patients was small, and confidence intervals for the individual estimation were wide, so the statistical precision of results is not ideal. Although the lower sensitivity means patients testing for TLS should use multiple diagnostic tests to rule out a negative result. PPV and NPV are always dependent on the underlying prevalence of the condition and may not be transferable to a new environment where the prevalence is different. This includes when the study inclusion/exclusion criteria might differ from usual clinical practice.

Conclusion

MRS is a type of noninvasive neuroradiology and a promising method for evaluating epilepsy patients since it has a higher sensitivity for detecting mesial temporal disease that is not seen on structural MRI imaging. As a result, EEG and clinical data, combined with MRI and MRS results, aid in identifying the epileptogenic zone in TLE patients. Furthermore, the detection of hippocampal sclerosis is linked to a positive surgical outcome for the patient, showing the technique's prognostic utility. The results of the study overall appear to be of interest but still need further support from future studies with larger sample sizes.

Funding Information

None.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request, subject to regulatory or other requirements.