Distinguishing autoimmune encephalitis from its mimics using brain [¹⁸F]-FDG PET: real-world study.

Abstract

Diagnosing autoimmune encephalitis (AE) remains challenging due to heterogeneous clinical presentations, frequent seronegativity, and limited MRI sensitivity. We evaluated the diagnostic performance of brain [¹⁸F]-FDG PET/CT in a real-world cohort of patients with suspected AE. Consecutive patients referred for [¹⁸F]-FDG PET/CT for suspected AE between January 2020 and July 2023 were retrospectively included. Final diagnoses were established using international criteria, expert consensus, and follow-up. PET findings were assessed visually, with voxel-wise and regional analyses, compared with 54 healthy controls, and stratified by age and treatment. Among 141 patients (48 ± 21 years), 33% had AE, 43% differential diagnoses (DD), and 24% remained undetermined. Of AE cases, 70% were definite, 63% seropositive, and 35% had supportive MRI findings. AE showed cortical hypometabolism with mesiotemporal lobe (MTL), basal ganglia, cerebellum, midbrain, and insula hypermetabolism (p < 0.05). Compared with DD, AE exhibited lower occipital/prefrontal and higher MTL, insular, and midbrain metabolism (p < 0.001), with good discrimination versus DD (AUC 0.82-0.83). MTL hypermetabolism, alone or with cortical hypometabolism, emerged as the most specific feature, whereas other regional metrics and ratios were less reliable (AUC ≤ 0.73). Accuracy of visual PET analysis was high in patients under 40 years (positive/negative predictive values 82%/81%) and declined with age to 81%/69% at 40-65 years and 79%/64% over 65 years. [¹⁸F]-FDG PET/CT identifies a characteristic metabolic signature of AE and differentiates it from relevant mimics, supporting its role as a complementary diagnostic biomarker, particularly in younger patients.