The purpose of this investigation was to determine the added value of metabolic imaging, perfusion mapping, and diffusion kurtosis imaging (DKI) in accurately identifying and forecasting outcomes for patients with H3K27-altered diffuse midline gliomas (DMGs). A total of 95 individuals (mean age 16.98 years; 61.1% female) with brainstem masses were enrolled between June 2020 and May 2023. This cohort included 71 patients with H3K27-altered DMGs and 24 patients harboring H3K27 wild-type brainstem tumors (labeled non-DMGs). All participants underwent pre-operative conventional MRI along with advanced sequences: amide proton transfer-weighted (APTw) imaging, arterial spin labeling (ASL), and DKI. The study employed logistic regression and Cox proportional hazards models with leave-one-out cross-validation (LOOCV) to examine the standalone and synergistic contributions of these advanced MRI parameters for both diagnostic classification and survival prediction. Incorporating advanced MRI parameters markedly strengthened the ability to diagnose and prognosticate H3K27-altered DMGs. APTw imaging proved particularly useful, demonstrating significant associations with diagnosis (odds ratio [OR] = 3.81, p = 0.004) and overall survival (Hazard ratio [HR] = 1.89, p = 0.003). ASL-derived relative cerebral blood flow (rCBF) also enhanced diagnostic accuracy (OR = 7.74, p = 0.019) and prognostic stratification (HR = 2.87, p = 0.002). Meanwhile, DKI-derived mean diffusivity (MD) was strongly linked to the presence of H3K27-altered DMGs (OR = 0.03, p = 0.012). Multivariate analysis showed that APTw (OR = 3.41, p = 0.014) and MD (OR = 0.05, p = 0.049) functioned as independent diagnostic indicators, while APTw (HR = 1.66, p = 0.037) and rCBF (HR = 2.15, p = 0.038) emerged as independent predictors of overall survival (OS). The combination of metabolic, perfusion, and advanced diffusion imaging techniques provides substantial improvements in the identification and outcome prediction of H3K27-altered DMGs compared to traditional MRI features. These insights have the potential to guide more effective clinical management strategies.
