Perfusion Parameter Obtained on 3-Tesla Magnetic Resonance Imaging and the Ki-67 Labeling Index Predict the Overall Survival of Glioblastoma.

BACKGROUND Pulsed arterial spin-labeling, diffusion tensor imaging (DTI), and magnetic resonance spectroscopy (MRS) are useful for predicting glioma survival. We performed a comparative review of multiple parameters obtained using these pulse sequences on 3-Tesla magnetic resonance imaging (MRI) including the molecular status and Ki-67 labeling index in newly diagnosed supratentorial glioblastomas. METHODS A total of 35 patients with glioblastomas underwent pulsed arterial spin-labeling, DTI, and MRS studies using 3-Tesla MRI preoperatively. The isocitrate dehydrogenase (IDH) mutation status, methylguanine-DNA methyltransferase methylation status, and Ki-67 labeling index were calculated from the tumor specimen. Cutoff values were identified by analyzing a receiver operating characteristic curve, and the multivariate survival statistical technique was performed to determine the significant and independent parameters for predicting overall survival. RESULTS The multivariate Cox analysis showed that the maximum/mean relative cerebral blood flow (rCBF) ratio and the Ki-67 labeling index were significant and independent predictive parameters with a cutoff value of 1.589 for the maximum rCBF ratio, 1.286 for the mean rCBF ratio, and 19% for the Ki-67 labeling index and hazard ratios of 6.132 and 5.119, respectively. The Kaplan-Meier survival curves showed that patients with higher rCBF ratios and Ki-67 labeling indices had a shorter overall survival than others, with median overall survival durations of 479 (95% CI, 370-559) and 1243 (95% CI, 666-NA) days, respectively (P = 0.000167). CONCLUSIONS Our findings indicate that the preoperative rCBF ratio and Ki-67 labeling index are useful parameters for predicting the overall survival of cerebral glioblastomas.