The aim of this study was to investigate the quantitative 18F-fluoroethylcholine (CHO) PET characteristics for differentiating lower-grade glioma (LGG) from glioblastoma (GBM).
Patients and Methods
Thirty-nine patients who underwent CHO PET with histopathologically confirmed primary diffuse glioma were prospectively enrolled. The 3-dimensional region of interest was semiautomatically defined based on the SUV threshold, and a total of 74 quantitative features, including 13 shape features, 31 SUV-based features, and 30 normalized SUV-based features, were calculated. Wilcoxon rank sum test, receiver operating characteristic curve, and correlation coefficient analyses were applied to select independent representative features, and patient prognosis was stratified by the World Health Organization (WHO) grade and CHO features.
A total of 89.2% of the quantitative features were significantly different between LGG and GBM, and the SUV-based features displayed higher area under the receiver operating characteristic curve (AUC) values than the other feature groups. Among the 5 traditional features, the SUVmax and the total lesion CHO uptake were the most distinguishing, with AUCs of 0.880 and 0.860 (0.938 and 0.927 after reclassification of 2 outliers), respectively, both of which could also stratify patient prognosis better than WHO grade. Five alternative features, including 2 shape features and 3 SUV-based features, were considered representative, with AUCs ranging from 0.754 to 0.854.
Quantitative features from CHO PET are reliable in determining the WHO grade of primary diffuse gliomas. Our findings suggest that GBM has a larger volume, a more spherical shape, higher choline activity in most interval segments, and a more symmetrical distribution than LGG.