Abstracts

Rationale:
Brain temperature mapping is a promising tool for localizing epileptogenic zones in MRI-negative patients with treatment-resistant epilepsy (TRE), especially in patients with suspected neuroinflammatory pathophysiology. Whole-brain (WB) magnetic resonance spectroscopic thermometry (MRSt) allows for noninvasive calculation of voxel-level brain temperature from the chemical shift difference between water and creatine (TempCRE = −102.61 × ΔH20-CRE + 206.1°C). Previous studies evaluated TempCRE using region-based approach, which averages temperature across voxels within 47 distinct ROIs to create a single integrated spectrum. The lack of findings using this approach, as well as its limited clinical utility, underscores the need to evaluate WB TempCRE maps in which voxels retain their original value. Our objective was to examine whether TempCRE abnormalities in epilepsy patients are evident in clinically-implicated brain regions, and whether the map- and region-based approaches are concordant.
Method:
Twenty healthy controls (HC, 22 - 46 YOA) and 9 epilepsy patients (22 - 49 YOA, 8 TRE) were scanned on a 3T Siemens Magnetom Prisma using a 20-channel head coil. T1-weighted structural images were acquired using a magnetization-prepared rapid gradient echo pulse sequence (224mm FOV, 320x320 matrix, TR=2400ms, TE=2.15ms, 0.7mm slice thickness). WB MRSt data were collected using a 3D echo planar spectroscopic imaging sequence (FOV of 280mm in-plane and 180mm in the inferior-superior direction to yield 5.6x5.6x14.4mm voxels, TR1=1710ms, TR2=591ms, TE=17.6ms, lipid inversion-recovery time of 198ms). Image reconstruction and spectral processing was completed within the Metabolite Imaging and Data Analysis System (MIDAS) software package. This included spatial reconstruction, frequency alignment, B0 inhomogeneity correction, lipid suppression, spectral fitting, and normalization and integration with water reference data. HCs’ TempCRE maps demonstrated a global TempCRE 37.0 ± 0.3°C (min 36.3°C - max 37.5°C). Thus, epilepsy patients’ TempCRE clusters ≥38°C (or ≥3 SD from the HC mean) were considered abnormal, with an upper threshold of 42°C (Fig. 1). For ROI-based TempCRE estimates, z-scores >1.65 SD (90% CI) from the HC mean were considered abnormal. Each patient’s map- and ROI-based TempCRE findings were assessed alongside clinical impression.  
Results:
The map- and ROI-based approaches for localizing TempCRE elevations differed greatly in their capacity to discriminate brain regions with subtle temperature elevations. TempCRE maps delineated abnormal temperature elevations in 9 out of 9 patients in accordance with clinical impression (Table 1). ROI-based results were limited, and only demonstrated TempCRE increases in 3 of 9 patients, with normal TempCRE estimates for all regions in the remaining participants (Table 1).
Conclusion:
MRSt-based TempCRE maps may allow mapping epileptogenic zone with more specific delineation of TempCRE elevations than the currently established ROI-based assessments. Future studies with larger sample size are necessary to better articulate concordance with clinical findings.
Funding:
:The State of Alabama General Funds supported this work.