Preoperative Prediction of H3 K27 Alteration Status in Diffuse Midline Glioma Using Intravoxel Incoherent Motion Imaging-based Habitat Imaging.

RATIONALE AND OBJECTIVES: The H3 K27 alteration status of diffuse midline gliomas (DMGs) is critical for prognosis and therapeutic decision-making but currently relies on invasive tissue sampling. MATERIALS AND METHODS: 153 patients (126 adults) with DMG who underwent intravoxel incoherent motion (IVIM) examinations were retrospectively included. DMGs were divided into H3 K27-altered (H3 K27-A) and wild-type (H3 K27-WT) groups. The DMGs were divided into three habitats (habitat 1-low diffusion and perfusion, habitat 2 -high diffusion and low perfusion, habitat 3-high perfusion) based on perfusion fraction and true diffusion coefficient maps. The quantitative parameters of each habitat were compared. The logistic regression was used to select risk factors. The clinical, habitat, and hybrid models were constructed using identified risk factors from clinical characteristics, habitat metrics, and their combination, respectively. The diagnostic performance and overall survival (OS) were analyzed. RESULTS: Compared with H3 K27-WT tumors, H3 K27-A DMGs had significantly higher f value in habitat 2 and habitat 3, and a larger volume of habitat 3 (all p < 0.05). The age, peritumoral nonenhanced pattern, and habitat metrics were selected as risk factors for constructing diagnostic models. The hybrid model achieved the highest diagnostic performance (area under the curve (AUC) = 0.900), followed by the clinical model (AUC = 0.860) and habitat model (AUC = 0.716). Only the predicted H3 K27-A group by the hybrid model showed significantly different OS compared with the H3 K27-WT group (p = 0.043). CONCLUSION: IVIM-based habitat imaging can noninvasively predict the H3 K27 alteration status of DMG with high accuracy, which is associated with prognosis.