45 Predicting sites of tumour progression in the invasive margin of glioblastomas (PRaM-GBM Study): a multicentre imaging study

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract Introduction Despite advances in surgery and radiotherapy, glioblastoma patients often experience progression within a year, adjacent to the resection cavity. Conventional imaging techniques are not sensitive enough to detect invasive tumour, leaving occult tumour undetected. We have developed a diffusion tissue signature technique using diffusion tensor MRI (DTI) that can locate occult tumour and predict their pattern of progression. This multicentre study aims to qualify DTI as a biomarker for tumour progression to improve surgical and radiotherapy treatment volumes. Method Patients with imaging finding of glioblastoma where the operating surgeon thought could resect >90%, were imaged preoperatively and at progression using diffusion tensor imaging (DTI). The DTI was processed to generate anisotropic diffusion tissue maps (DTI-q). Both sets of images were coregistered, and each pre-operative voxel was classified based on the imaging at progression on a voxel-by-voxel basis. Voxels were classified as true positives, true negatives, false positives, and false negatives. This allowed calculation of the sensitivity and specificity for each patient. These were combined, and confidence intervals were calculated based on 1000 bootstrap samples. Results In total, 139 patients were recruited from 5 neurosurgical units. 54 patients had to be withdrawn, leaving 85 evaluable patients. 82% of patients underwent complete resection of the enhancing tumour. Overall, the sensitivity of DTI-q to predict sites of tumour progression was 79.3% (95% CI: 72.3-85.4%), and the specificity was 93.1% (92.3-93.9%). Conclusion This study has shown that diffusion tissue signatures can identify sites of tumour progression with high sensitivity and specificity. Future studies to change the resection and the radiotherapy treatment volumes are underway. This may allow us to personalise treatment volumes for glioblastomas.