ObjectivesTo comprehensively and noninvasively predict glioma grade, IDH mutation status, 1p/19q codeletion status, and MGMT promoter methylation status using chemical exchange saturation transfer (CEST)-based tumor pH assessment and metabolic profiling.MethodsWe analyzed 128 patients with pathologically confirmed adult diffuse glioma. CEST-derived metrics based on tumor regions were obtained using five-pool Lorentzian analysis and pH_weighted analysis. Histogram features of these metrics were computed to characterize tumor heterogeneity. These features were subsequently employed for glioma grading and molecular genotyping of IDH, 1p/19q and MGMT. Logistic regression analysis was used to predict the grade and IDH genotypes. The diagnostic performance was evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC) analysis.ResultsThe DS, MT and pH_weighted differed significantly between grade II and III, as well as grade III and IV. The amide, NOE, pH_weighted and MTR3.5 showed significantly differences within IDH genotypes. Regression models achieved the highest AUC for differentiating grade II from III (0.80, 95% CI: 0.64-0.91), grade III from IV (0.83, 95% CI: 0.74-0.90), and IDH mutant from wild status (0.84, 95% CI: 0.77-0.90). MT and pH_weighted metrics were the only indicators for identifying 1p/19q codeletion in grade II and grade III gliomas, respectively. MT 90th percentile (0.87, 95% CI: 0.65-0.98) and pH_weighted 25th percentile (0.83, 95% CI: 0.56-0.97) showed the best performance, respectively. The MTR3.5 was the only indicator which can distinguish MGMT promoter methylation and unmethylation gliomas, within MTR3.5 90th percentile performed best (AUC = 0.79, 95% CI: 0.61- 0.91).ConclusionCEST-based tumor pH assessment and metabolic profiling demonstrated promising potential for predicting glioma grade, IDH mutation status, 1p/19q codeletion, and MGMT genotype.
