Quantitative dual-tracer 18F-FDG and 18F-DPA-714 PET/MR for characterizing pathological subtypes in refractory mesial temporal lobe epilepsy.

BACKGROUND: Pathological subtypes are key determinants of surgical strategy and prognosis in refractory mesial temporal lobe epilepsy (MTLE), with hippocampal sclerosis (HS) associated with better outcomes than gliosis only. Subtype-specific patterns of cerebral metabolism and neuroinflammation, which may inform lesion localization and subtype differentiation, remain poorly defined. This study aimed to delineate these patterns using ¹⁸F-FDG PET and translocator protein (TSPO) PET with ¹⁸F-DPA-714 for precise localization, with emphasis on the added value of quantitative analysis. METHODS: Patients with unilateral refractory MTLE undergoing surgery were retrospectively included, and pathological subtypes were classified according to histological profiles (HS, n = 43; gliosis only, n = 41). Patients underwent sequential integrated dual-tracer PET/MR imaging within one week. Age- and sex-matched healthy controls (HC) were recruited. Whole-brain voxel-wise and temporal lobe subregional analyses were performed. Localization performance was assessed by visual inspection and quantitative analysis employing an asymmetry index (AI) derived from standardized uptake value ratios. Sensitivity and accuracy were calculated, with false discovery rate correction applied for multiple comparisons. RESULTS: In total, 43 MTLE-HS and 41 MTLE-gliosis only were included. Compared with HC, both HS and gliosis exhibited significant 18F-FDG hypometabolism alongside increased 18F-DPA-714 uptake, more pronounced in HS. Whole-brain 18F-DPA-714 abnormalities were more confined to temporal regions, affording clearer lateralization and localization than 18F-FDG. In temporal-lobe subregional analysis, 18F-FDG outperformed 18F-DPA-714 for subtype discrimination, and hippocampal AI value achieved the best performance (35/43 [81.4%] vs. 22/41 [53.7%], p = 0.007). Quantitative analysis enhanced localization accuracy for both tracers, yielding the greatest improvement for TSPO in gliosis (from 10/41 [24.4%] to 19/41 [46.3%], p = 0.038). CONCLUSION: Dual-tracer PET/MR combined with quantitative analysis may improve subtype-specific presurgical localization in refractory MTLE. TSPO-PET precisely localized the epileptogenic lesion, whereas FDG-PET facilitated differentiation of pathological subtypes.