Diffusion tensor invasive phenotypes can predict progression-free survival in glioblastomas.


INTRODUCTION: Glioblastomas multiformes (GBM) remain incurable in most cases. Their invasion into normal brain makes current therapies ineffective. Post-mortem studies suggest about a 25% of GBMs invade less than 1 cm from the tumour bulk and 20% invade more than 3 cm. AIM OF STUDY: The study aims to use DTI to assess tumour extension and determine how previously reported patterns relate to the progression-free survival (PFS). MATERIALS AND METHODS: Twenty-five patients with GBM treated according to the EORTC/NCIC protocol were retrospectively analysed. Patients were imaged post-operatively at 1.5 T. The sequences were composed of standard anatomical and a standard DTI sequence. As described earlier p and q maps were constructed. For each of the p and q maps, regions of interest were drawn around the visible abnormality. Patients were assigned a diffuse, localised or minimally invasive pattern. Progression was defined according to the RANO criteria (4) and PFS determined in days. Kaplan-Meier plots of survival for the three groups were plotted as were the proportion of patients who had not progressed at 24 months. RESULTS: The median PFS for the diffuse group was 278 days, for the localised group 605 days and 820 days for the minimally invasive group. Three-fourth of the minimally invasive group were progression-free at 24 months (LOG RANK 9.25; p = 0.010). CONCLUSION: It is possible to identify three invasive phenotypes in GBMs using Diffusion tensor imaging , and these three phenotypes have different progression free survival. A minimal phenotype (20% of patients) predicts a greater delay to progression.