- publication
- 13-03-2025
BackgroundExtent of resection (EOR), including gross total resection (GTR), is one of the most important factors in predicting overall survival (OS) in IDH-wild type (IDH-WT) glioblastoma patients. Although GTR represents the complete resection of all visible contrast-enhancing parts of the tumor, imaging predictors of achieving this extent still need to be better understood.PurposeTo assess the impact of preoperative imaging phenotypes as defined by the VASARI feature set and tumoral volumetry to determine predictors of GTR in patients with IDH-WT glioblastoma.Material and MethodsThis retrospective, single-center study analyzed imaging characteristics based on the VASARI features in the preoperative scans of IDH-WT glioblastoma patients. Volumetric analysis was performed to determine associations with clinical outcomes. Univariate analysis was used to determine the association of VASARI features with GTR. A multivariate analysis model was used to determine predictors of GTR.ResultsGTR was achieved in 79/144 (54.8%) patients, near total resection in 15 (10.4%), and subtotal resection in 50 (34.7%) patients. Our results showed non-eloquent tumor regions (55% vs. 35%; = 0.04) and thick margin of enhancement (56.1% vs. 43.9%; = 0.04) were associated with GTR and ependymal extension (37% vs. 63%; = 0.02). Deep white matter invasion (36.3% vs. 63.7%; = 0.03) was significantly associated with non-gross total resection. Lower tumoral volumes were also associated with gross total resection (< 0.01). After performing multivariate analysis, the thickness of the tumoral enhancing margins was correlated with GTR with an OR of 1.57 (95% CI=1.1-2.23). Furthermore, the volume of the enhancing component was significantly different according to EOR with a calculated OR of 0.95 (95% CI = 0.92-0.97; < 0.01).ConclusionImaging characteristics on standard-of-care MRI can predict the rate of GTR in patients with IDH-WT glioblastomas. The thickness of enhancing margins predicts GTR after multivariate analysis. A diagnostic model that includes a combination of the discriminating depicted features on MRI and brain tumor volumetrics has an acceptable diagnostic performance with a specificity >90%.
voir sur Pubmed