Noncontrast Magnetic Resonance Fingerprinting-Habitat Framework for Intratumoral Heterogeneity Decoding and Isocitrate Dehydrogenase Mutation Prediction in Diffuse Gliomas.

PURPOSE: To develop and validate a magnetic resonance fingerprinting (MRF)-based habitat imaging framework for noninvasively decoding intratumoral heterogeneity (ITH) and preoperatively predicting isocitrate dehydrogenase (IDH) mutation status in diffuse gliomas.

MATERIALS AND

METHODS

In this prospective study (January 2024-September 2025), 141 adults with diffuse gliomas (56 IDH-mutant, 85 IDH-wildtype) were enrolled. Tumors were segmented into three habitats via K-means clustering of coregistered MRF-derived T2 and free water maps. A habitat-based radiomic model for IDH status was developed in a training cohort (n = 98) and validated in an independent test cohort (n = 43). Its performance was compared against a conventional whole-tumor model using area under the curve (AUC) and net reclassification improvement (NRI). Pathophysiological validation was performed by correlating habitats with the Ki-67 proliferation index, dynamic contrast-enhanced magnetic resonance imaging (Ktrans), and apparent diffusion coefficient (ADC). The prognostic value for progression-free survival (PFS) was also assessed.

RESULTS

The MRF-habitat model outperformed the whole-tumor model for IDH genotyping (test AUC, 0.819 v 0.758; NRI, 0.833, P = .003). It identified a prognostically significant subregion (Subregion1) where T2 uniformity was an independent predictor of PFS (hazard ratio: 0.008; P = .033). Subregion2, a hypoxic-angiogenic niche, correlated with Ki-67 and exhibited higher Ktrans and lower ADC in IDH-wildtype tumors (both P < .05). Patients stratified as high-risk by the model had significantly shorter median PFS (5.8 months v not reached, P = .043).

CONCLUSION

The MRF-based habitat framework noninvasively decodes ITH, improves preoperative IDH genotyping, and identifies pathophysiologically distinct subregions with prognostic relevance.