Identification of early response to anti-angiogenic therapy in recurrent glioblastoma: Amide proton transfer–weighted and perfusion-weighted MRI compared with diffusion-weighted MRI

Background: Amide proton transfer (APT) MRI has the potential to demonstrate antitumor effects by reflecting biologically active tumor portion, providing different information from diffusion-weighted imaging (DWI) or dynamic susceptibility contrast (DSC) imaging. Purpose: To evaluate whether a change in APT signal intensity after antiangiogenic treatment is predictive of early treatment response in recurrent glioblastoma. Materials and Methods: In this retrospective study, APT MRI, DWI, and DSC imaging were performed in patients with recurrent glioblastoma from July 2015 to April 2019, both before treatment and 4–6 weeks after initiation of bevacizumab (follow-up). Progression was based on pathologic confirmation or clinical-radiologic assessment, and progression patterns were defined as local enhancing or diffuse nonenhancing. Changes in mean and histogram parameters (fifth and 95th percentiles) of APT signal intensity, apparent diffusion coefficient, and normalized cerebral blood volume (CBV) between imaging time points were calculated. Predictors of 12-month progression and progression-free survival (PFS) were determined by using logistic regression and Cox proportional hazard modeling and according to progression type. Results: A total of 54 patients were included (median age, 56 years [interquartile range, 49–64 years]; 24 men). Mean APT signal intensity change after bevacizumab treatment indicated a low 12-month progression rate (odds ratio [OR], 0.36; 95% confidence interval [CI]: 0.13, 0.90; P = .04) and longer PFS (hazard ratio: 0.38; 95% CI: 0.20, 0.74; P = .004). High mean normalized CBV at follow-up was associated with a high 12-month progression rate (OR, 20; 95% CI: 2.7, 32; P = .04) and shorter PFS (hazard ratio, 9.4; 95% CI: 2.3, 38; P = .002). Mean APT signal intensity change was a significant predictor of diffuse nonenhancing progression (OR, 0.27; 95% CI: 0.06, 0.85; P = .047), whereas follow-up 95th percentile of the normalized CBV was a predictor of local enhancing progression (OR, 7.1; 95% CI: 2.4, 15; P = .04). Conclusion: Early reduction in mean amide proton transfer signal intensity at 4–6 weeks after initiation of antiangiogenic treatment was predictive of a better response at 12 months and longer progression-free survival in patients with recurrent glioblastoma, especially in those with diffuse nonenhancing progression.