Identifying the survival subtypes of glioblastoma by quantitative volumetric analysis of MRI.

Abstract

This study was to project a powerful volumetric-related parameter on magnetic resonance imaging (MRI) for classifying patients with glioblastoma multiforme (GBM) into distinct subgroups objectively. The preoperative MRIs of 147 patients with primary GBM were analyzed. Volumetric-related parameters, including V1 (tumor volume), V2 (peritumoral T2/FLAIR hyperintense volume) and V2/V1 (the volume ratio), were estimated by an ellipsoid model. Log-rank analysis and Cox regression methods were used to compare Kaplan-Meier plots and identified prognostic parameters. Log-rank analysis revealed that V1 and V2 were correlated with survival, but the P value was marginally significant (P = 0.082, P = 0.091, for progression-free survival [PFS]; P = 0.120, P = 0.073, for overall survival [OS], respectively). V2/V1 was a potential prognostic factor for both PFS and OS (P < 0.001 and P < 0.001, respectively). Cox regression analysis documented that higher V2/V1 (ratio ≥ 7.0) was independent unfavorable prognostic factor. The odd ratio (OR) of higher V2/V1 was 2.662 (95 % confidence interval [CI], 1.782-3.975; P < 0.001) for PFS and 3.450 (95 % CI, 2.079-5.725; P < 0.001) for OS, respectively. The volumetric-related parameters of V1, V2 and V2/V1 were helpful for predicting the prognosis of patients with GBM. V2/V1 was a more comprehensive and systematic prognostic factor in GBM patient, especially for those with small tumor but large peritumoral T2 hyperintense or large tumor but small peritumoral T2 hyperintense.