INTRODUCTION
Epilepsy surgery is an effective therapy for drug-resistant epilepsy^1,2, and surgery is more successful in patients with epileptogenic lesions detected with structural MRI. About 20-30% of patients with focal epilepsy show no structural abnormalities, i.e., are MRI-negative³. Surgery can be effective if an epileptic focus can be defined in another way. Metabolic MRI can visualize underlying metabolic pathways associated with disease, especially when applied at ultrahigh field with its increased SNR and spectral dispersion to help us identify epileptogenic lesions not detected with structural MRI.

Several potential metabolic markers for epileptogenic lesions that can be imaged with MRI have been reported in recent years. For instance, glutamate, associated with mitochondrial and metabolic injury inducing seizures in patients with epilepsy, can be measured with MRS, and locally increased concentrations have been found in MRI-negative epileptogenic lesions.⁴ Also iron accumulation, a potential marker of oxidative stress, has been found in hippocampal sclerosis (HS) and focal cortical dysplasia (FCD) using histopathological staining.⁵ The metabolic phenotype – or ‘profile’ – of epileptogenic lesions will likely consist of a combination of metabolites, some of which may not even have been found using in vivo imaging techniques.

In this preliminary clinical trial, we aim to find (combinations of) metabolic markers of epileptogenic lesions that can be detected with 7T metabolic MRI, using glutamate-CEST (GluCEST), quantitative susceptibility mapping (QSM) and single-voxel (SV) ¹H MRS, and validate these with histopathology. This abstract discusses the imaging results of QSM and ¹H MRS in the first five patients compared to five healthy volunteers.

METHODS
Five patients (aged 23-48 years; three females) and five healthy volunteers (aged 24-27 years; two females) were included. All patients were diagnosed with focal epilepsy with an epileptogenic lesion on structural MRI and/or positron emission tomography and were scheduled for surgical treatment. This clinical trial was approved by our IRB; all participants gave written informed consent.

MRI examination
All subjects underwent one 7T (Philips, Best, the Netherlands) MRI examination with a 32-channel head coil (Nova Medical) the day prior to their surgery. Next to QSM and SV ¹H MRS (sLASER), a high-resolution T₁-weighted sequence was acquired for FOV planning; see Table 1 for acquisition parameters. The FOV in healthy volunteers was matched with the lesion location in the patients; for MRS, a TE of 36 ms was used in HS patients (N=3) and matched volunteers, and a TE of 110 ms was used in the FCD patients (N=2) and matched volunteers. Final diagnosis was based on histopathology.

Post-processing & data analysis
QSM data was post-processed with the SEPIA tool using Laplacian-based phase unwrapping, variable-kernel SHARP background field removal and the iLSQR method for computing the susceptibility maps. MRS data was analyzed using LCModel with a simulated basis set.⁶ No correction for T₁ or T₂ relaxation times was applied. Metabolite ratios with respect to total creatine, with a Cramér Rao Lower Bound of <20% for N≥2 per different TE were reported (Figure 1; Table 2). QSM images were analyzed qualitatively for differences in iron deposition; a Wilcoxon-signed rank test was used to compare differences in metabolite ratios between epileptic lesions and healthy brain tissue.

RESULTS
QSM
The two patients with histopathologically confirmed HS showed increased iron deposition in the affected hippocampus, while the third HS patient did not show abnormal iron deposition in either hippocampus (Figure 2); this patient however turned out not to have abnormal tissue. None of the matched healthy volunteers nor the patients with FCD showed increased iron deposition in either hippocampus or respective lesions of interest.

¹H MRS
Some differences in metabolite ratios between HS and the contralateral hippocampus in the two histopathologically confirmed HS patients could be seen, though not statistically evident. We also found no evident differences in metabolite ratios between HS patients and healthy volunteers. This same result was found in the FCD patients and their matched healthy volunteers.

DISCUSSION & CONCLUSION
In this clinical study, we combined different (metabolic) MRI sequences at 7T to characterize focal epileptogenic lesions, and uncover potential metabolic markers that could help identifying the culprit lesion in MRI-negative epilepsy patients. Using QSM, we observed increased iron deposition in the affected hippocampus of two HS patients that was not found in the contralateral hippocampus, nor in a patient with suspected HS with no abnormal tissue, nor in matched healthy volunteers. These results support previous findings of increased iron deposition using susceptibility-weighted imaging and histopathology in temporal lobe epilepsy (TLE)^5,7; however, contrary to previous histopathology-based literature, no increased iron deposition was found in patients with FCD. Two studies at 7T using MRS in epilepsy patients showed variable metabolite ratios: both increased (FCD) and ‘normal’ (TLE) GABA levels were found, and no differences in glutathione and glutamate/glutamine.^8,9 The same lack of statistically significantly different metabolite ratios was found in our five patients; however, our very small sample size precludes any final conclusions. Ongoing patient inclusion will increase our sample size, allowing for more robust statistical analyses of our QSM and MRS data. Combining these with planned GluCEST data acquisition in the same patients could lead to an even stronger profiling of the metabolic phenotype of epileptogenic lesions.

Acknowledgements

We would like to thank neurosurgeons dr. P.C. van Rijen, drs. P.H. Gosselaar and nurse specialist J.M. Ophorst-van Eck from the University Medical Center Utrecht for their contribution to this study.