Glioma recurrence versus radiation necrosis? A pilot comparison of arterial spin-labeled, dynamic susceptibility contrast enhanced MRI, and FDG-PET imaging.

摘要

RATIONALE AND OBJECTIVES: Distinguishing recurrent glial tumor from radiation necrosis can be challenging. The purpose of this pilot study was to preliminarily compare unenhanced arterial spin-labeled (ASL) imaging, dynamic susceptibility contrast-enhanced cerebral blood volume (DSCE-CBV) magnetic resonance imaging, and positron emission tomographic (PET) imaging in distinguishing predominant glioma recurrence or progression from predominant radiation necrosis in postoperative patients treated with proton-beam therapy. METHODS: Patients with grade II to IV glioma previously treated with surgery and proton-beam therapy were enrolled on the basis of new enhancing nodules or masses with primary differential diagnoses of predominant tumor recurrence or progression versus radiation necrosis. ASL, DSCE-CBV, and PET examinations were assessed by visual qualitative and quantitative analysis for the detection of predominant tumor recurrence. Imaging results were correlated with a clinical-pathologic reference standard. RESULTS: Thirty patients were studied, resulting in 33 ASL, 32 DSCE-CBV, and 26 PET examinations. On the basis of visual inspection, the sensitivities of PET, ASL, and DSCE-CBV examinations for detecting high-grade tumor foci were 81%, 88%, and 86%, respectively. The highest sensitivity values for quantitative ASL imaging were obtained using a normalized cutoff ratio of 1.3, resulting in sensitivity of 94% for ASL imaging and 71% for DSCE-CBV imaging. When predominant high-grade tumors with superimposed regions of predominant mixed radiation necrosis were excluded, DSCE-CBV sensitivity improved to 90%, but ASL sensitivity remained unchanged. CONCLUSIONS: Compared with DSCE-CBV imaging, ASL imaging may more accurately distinguish predominant recurrent high-grade glioma from radiation necrosis, especially in regions with mixed radiation necrosis, for which DSCE-CBV imaging may underestimate true blood volume because of leakage artifacts.