Multiscale DCE-BOLD MRI reveals vascular uncoupling patterns for IDH genotyping in adult glioma: A prospective habitat-based study

Isocitrate dehydrogenase (IDH)-wildtype gliomas exhibit heterogeneous vascular dysfunction, but whole-tumor analyses may obscure regional variation in vascular permeability and functional reserve. We developed a multiscale habitat framework integrating dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI)-derived volume transfer constant (K^trans) and resting-state blood oxygen level-dependent functional MRI (rs-BOLD-fMRI)-derived relative cerebrovascular reactivity (rCVR), including a 10-mm peritumoral margin, for noninvasive IDH genotyping. This prospective single-center cohort included 132 patients (52 IDH-mutant and 80 IDH-wildtype), divided chronologically into training (n = 92) and internal test (n = 40) cohorts. Gaussian mixture modeling partitioned the continuous joint K^trans-rCVR distribution into three overlapping imaging-defined intratumoral subregions; the anatomically defined peritumoral region was analyzed separately. The Habitat-Peri10mm model yielded the highest internal test area under the receiver operating characteristic curve (AUC) point estimate of 0.914 (95% confidence interval, 0.831-0.997), compared with 0.854 for the Habitat model and 0.753 for the Whole-tumor model. Pairwise DeLong tests were nonsignificant (p = 0.173 and 0.054, respectively). A Subregion 2 K^trans first-order Entropy feature ranked highest in the SHapley Additive exPlanations (SHAP) analysis and correlated with Ki-67 (r = 0.57) and microvessel density (r = 0.27). The habitat-derived intratumoral heterogeneity score was higher in IDH-wildtype than IDH-mutant tumors (0.79 vs. 0.40, p < 0.001; AUC = 0.824), although grade imbalance limited IDH-specific interpretation. This DCE-BOLD multiscale habitat framework characterizes IDH-associated patterns of vascular heterogeneity and shows potential for noninvasive IDH genotyping.