SURG-20. PREOPERATIVE QUANTITATION OF UNRESECTABLE TUMOR VOLUME USING WHOLE-BRAIN TRACTOGRAPHY TO IMPROVE OUTCOME PREDICTION AFTER SURGERY FOR HIGH-GRADE DIFFUSE GLIOMAS

Abstract BACKGROUND Because minimizing injury to critical brain regions during high-grade glioma (HGG) surgery is essential, the optimal extent of resection (EOR) for any given patient is often less than 100% of the tumor bulk. However, there are no accepted methods of assessing the upper limit of EOR. Advances in whole-brain tractography (WBT) present an opportunity to address this problem. Here, we develop a predictive model for surgical outcomes in HGG as classified by WHO CNS5, and leverage it to determine the added utility of a WBT-based volumetric resectability metric. METHODS We included patients who underwent biopsy or resection of a new or recurrent HGG by the senior author from 2013 to 2020. A volumetric tumor analysis method incorporating WBT was developed to calculate EOR and unresectable tumor volume (UTV). Using perioperative clinical data, multiple linear and Cox regression models were used to identify independent predictors of EOR and progression-free survival (PFS). RESULTS A total of 151 HGGs were included. Median EOR and PFS were 97.0% and 5.0 months, respectively. Greater EOR strongly predicted longer PFS overall (hazard ratio [HR]=0.21, 95% confidence interval [CI] 0.07-0.72), with the largest PFS benefit observed at > 85% EOR. When controlling for EOR, higher residual tumor volume (RTV) predicted longer PFS (HR=0.97, 95% CI 0.94-0.99), suggesting the existence of an RTV threshold below which PFS decreases. Involvement of eloquent regions (p=0.0086) and brainstem (p=0.0006) as categorical variables predicted lower EOR but failed to predict shorter PFS. Calculation of UTV based on actual or imputed WBT data was feasible and reliably predicted EOR. CONCLUSION Existing methods of measuring HGG resectability are poorly reproducible and do not independently predict survival. Preoperative measurement of WBT-based UTV, by substituting for EOR, may simplify and improve HGG outcome prediction, allowing neurosurgeons to better assess the surgical candidacy of individual patients.