Advanced imaging characterization of post-chemoradiation glioblastoma stratified by diffusion MRI phenotypes known to predict favorable anti-VEGF response

Abstract

Purpose: Recurrent glioblastomas showing a survival benefit from anti-VEGF agents are known to exhibit a distinct diffusion MRI phenotype. We aim to characterize advanced imaging features of this glioblastoma subset. Methods: MRI scans from 87 patients with IDH-wildtype glioblastoma were analyzed. All patients had completed standard chemoradiation and were anti-VEGF-naïve. Contrast-enhancing tumor segmentations were used to extract: the lowest peak of the double gaussian distribution of apparent diffusion coefficient values (ADCL) calculated from diffusion MRI, relative cerebral blood flow (rCBV) values from perfusion MRI, MTRasym @ 3ppm from pH-weighted amine CEST MRI, quantitative T2 and T2* relaxation times (qT2 and qT2*), T1w subtraction map values, and contrast-enhancing tumor volume. Lesions were categorized as high- or low-ADCL using a cutoff of 1240 µm2/s, according to previous studies. Results: High-ADCL lesions showed significantly lower rCBV (1.02 vs. 1.28, p = 0.0057), higher MTRasym @ 3ppm (2.36% vs. 2.10%, p = 0.0043), and higher qT2 (114.8 ms vs. 100.9 ms, p = 0.0094), compared to low-ADCL lesions. No group differences were seen in contrast-enhancing tumor volume, T1w subtraction map values, and qT2*, nor in clinical variables such as sex category, MGMT status, and EGFR status. Finally, no clear group-specific preferential locations were seen. Conclusion: Post-chemoradiation glioblastomas with a diffusion MRI phenotype that is known to predict a favorable response to anti-VEGF (ADCL ≥1240 µm2/s) have distinct biological features, with different perfusion and metabolic characteristics, and T2 relaxation times.