EXTENT OF RESECTION IN GLIOBLASTOMA: REFINEMENT AND PROGNOSTIC VALIDATION OF A CLASSIFICATION SYSTEM FROM THE RANORESECTGROUP

Abstract Background Extent of resection in glioblastoma is inconsistently described across clinical trials. Based upon the absolute residual contrast-enhancing (CE) tumor (in cm3) and the relative reduction of CE tumor (in percentage) on postoperative MRI, a surgical classification system for glioblastoma was previously proposed. In this context, class 0 was defined as ‘supramaximal CE resection’ (also including removal of non-CE tumor), class 1 as ‘maximal CE resection’, class 2 as ‘submaximal CE resection’, and class 3 as ‘biopsy’. We herein aimed to (I) analyze the prognostic value of the proposed classification system, and (II) explore how much non-CE tumor needs to be resected in order to provide a survival benefit. Material and Methods An international Response Assessment in Neuro-Oncology (RANO) group was formed, entitled RANO resect. The RANO resectinvestigators retrospectively searched the databases from seven neuro-oncological centers in the USA and Europe for individuals with newly diagnosed glioblastoma. Demographics, clinical information and volumetrics from pre- and postoperative MRI were collected. Kaplan-Meier survival analysis and log-rank test were utilized to calculate survival, and Cox’s proportional hazard regression model was used to adjust for multiple variables. The significance level was set at p ≤ 0.05. Results We identified 1021 individuals with newly diagnosed glioblastoma, including 1008 IDHwt patients. Among those, 744 IDHwt patients were postoperatively treated with radiochemotherapy per EORTC 26981/22981 (TMZ/RT→TMZ). Within this homogenously treated cohort, higher extent of resection was favorably associated with outcome: individuals with ‘maximal CE resection’ (class 1) had superior outcome compared to patients with ‘submaximal CE resection’ (class 2) or ‘biopsy’ (class 3) (median OS: 20 versus 16 versus 10 months; p = 0.001). Similar findings were made when assessing progression (median PFS: 9 versus 8 versus 5 months; p = 0.001). Extensive resection of non-CE tumor (≥60% of non-CE tumor removed and ≤5 cm3 residual non-CE tumor) provided an additional survival benefit in patients with complete CE resection (class 1), thus defining class 0 (‘supramaximal CE resection’) (median OS: 29 versus 20 months; p = 0.003). The favourable prognostic effect of CE resection was retained in a multivariate analysis when stratifying for clinical and molecular characteristics including pre-operative tumor volume and MGMT promotor status (p = 0.001). Conclusion The proposed classification system to describe extent of glioblastoma removal is highly prognostic, and thus may serve for stratification of clinical trials. Resection of non-CE tumor beyond the CE tumor borders provides an additional survival benefit in glioblastomas, highlighting the need to explicitly denominate such a ‘supramaximal CE resection’.