Dynamic susceptibility contrast perfusion-weighted imaging (DSC-PWI) vascular and metabolic biomarkers for presurgical diagnosis of adult brain tumors across different acquisition parameters.

BACKGROUND: The diagnostic performance of dynamic-susceptibility-contrast perfusion-weighted-imaging (DSC-PWI) relies almost exclusively on cerebral-blood-volume (CBV), a biomarker considered highly dependent on acquisition parameters. We investigated whether integrating advanced vascular and metabolic perfusion biomarkers improves the presurgical diagnosis of brain tumors and whether these gains are maintained independently of acquisition parameters. METHODS: Retrospective single-center study (2023-2025) including 219 patients (101 glioblastomas, 52 metastases, 46 meningiomas, 20l ymphomas) with a unique double acquisition, single-session presurgical DSC-PWI protocol (first, non-preloaded high-flip-angle, and second, preloaded intermediate-flip-angle, per patient). Using innovative clinical software, eight normalized metrics were extracted from three-dimensional tumor masks. Three macrovascular: CBV, cerebral-blood-flow (CBF), mean-transit-time (MTT). Three microvascular: time-to-drain (TTD), capillary-transit-time-heterogeneity (CTH), coefficient-of-variation-of-transit-times (COV). Two metabolic: oxygen-extraction-fraction (OEF), cerebral-metabolic-rate-of-oxygen (CMRO2). Performance was assessed by the area-under-the-receiver-operating-characteristic-curve (AUC-ROC) in three acquisition scenarios: two homogeneous and one simulated heterogeneous. Independent validation was performed using a historical cohort of 55 unseen patients (2015-2020). RESULTS: Median AUC-ROC was 0.88 (0.87-0.90, without significant differences across scenarios). Specifically, the median AUC-ROCs per pair were: Lymphoma-Glioblastoma = 0.90, Metastasis-Glioblastoma = 0.79, Glioblastoma-Meningioma = 0.91, Lymphoma-Metastasis = 0.90, Metastasis-Meningioma = 0.82, Lymphoma-Meningioma = 0.97. These results outperformed the conventional cerebral blood volume (median 0.73, range 0.58-0.91; significant, paired Wilcoxon p < 0.001-0.04). Performance was preserved in the independent validation cohort despite its inherent real-world heterogeneity (median 0.82-0.85; individual comparisons 0.79-0.98). CONCLUSIONS: Advanced vascular and metabolic perfusion biomarkers enabled multidimensional tumor profiling beyond conventional CBV, achieving high and reproducible diagnostic performance across acquisition parameters and providing a potential framework for robust presurgical DSC-PWI assessment of the main adult brain tumors.