Abstract / Summary
Abstract Background Glioma grading has traditionally relied on histopathological assessment, while conventional MRI provides important but sometimes nonspecific information regarding tumor characteristics. Diffusion tensor imaging (DTI) provides quantitative measures of tissue microstructural organization, particularly fractional anisotropy (FA) and mean diffusivity (MD), which may provide additional information regarding glioma grade. Objective To evaluate the association between DTI parameters and histopathological glioma grade and to assess the relationship of DTI parameters with Ki-67 expression and their exploratory diagnostic performance in differentiating Grade 1–2 from Grade 3–4 tumors. Materials and methods This prospective observational study included 33 patients with histopathologically confirmed glioma who underwent preoperative MRI with DTI. MRI was performed at 1.5 T in 10 patients and 3 T in 23 patients. FA and MD were measured using manually placed ROIs in the tumor core, peritumoral region and contralateral normal-appearing brain parenchyma. Differences across tumor grades were assessed using the Kruskal–Wallis test with Dunn–Bonferroni post-hoc analysis. Spearman correlation was used to evaluate associations between DTI parameters, tumor grade and Ki-67 expression. Exploratory ROC analysis was performed for differentiation of Grades 1–2 from Grades 3–4. Results Tumor FA demonstrated a significant inverse correlation with glioma grade (Spearman ρ = −0.694, p < 0.001) and a significant difference across tumor grades (Kruskal–Wallis p = 0.001). Tumor MD did not demonstrate a significant correlation with grade (ρ = 0.258, p = 0.147). Peritumoral FA demonstrated a significant inverse correlation with grade (ρ = −0.793, p < 0.001). Tumor FA also showed a significant inverse correlation with Ki-67 expression (ρ = −0.632, p < 0.001), whereas tumor MD did not (ρ = 0.259, p = 0.145). On exploratory ROC analysis, tumor FA demonstrated an AUC of 0.833 (95% CI 0.688–0.952) for differentiating Grades 1–2 from Grades 3–4. Conclusion Tumor FA demonstrated a significant inverse association with glioma grade, whereas tumor MD did not show a significant grade-related association. The inverse association between FA and Ki-67 further supports a relationship between FA and tumor biological characteristics in this cohort. However, the small and heterogeneous cohort, variation in MRI acquisition, incomplete molecular profiling and absence of independent validation limit immediate clinical application. Larger multicenter studies with standardized DTI protocols and comprehensive molecular characterization are warranted.