Prognostic evaluation of surgical re-resection for recurrent glioblastoma using the novel RANO classification for extent of resection: A report of the RANO resect group

2010 Background: The clinical effects of re-resection for recurrent glioblastoma remain controversial, and the role of post-operative tumor volume is unclear since leaving certain tumor volumes deliberately behind cannot be ethically justified. A surgical classification system was previously proposed for stratification based upon residual contrast-enhancing (CE) tumor: RANO class 1 was defined as ‘supramaximal CE resection’ (including non-CE tumor removal), class 2 as ‘maximal CE resection’, class 3 as ‘submaximal CE resection’, and class 4 as ‘biopsy’. We aimed to (I) explore the prognostic role of extent of re-resection using this classification and (II) define clinical factors which consolidate the effects of re-resection. Methods: The RANO resect group retrospectively compiled a global, eight-center cohort of patients with first recurrence from a previously resected glioblastoma. The combined associations of re-resection and clinical factors with outcome were analyzed. Kaplan-Meier survival analysis and log-rank test were applied to calculate survival, and Cox’s proportional hazard regression models to adjust for multiple variables (significance level: p ≤ 0.05). A propensity score-matched analysis was constructed to mimic a randomized clinical trial comparing the different RANO classes. Results: We encountered 681 patients with first recurrence of IDH-wildtype glioblastoma, including 310 patients who underwent re-resection at first recurrence. The use of re-resection was associated with prolonged survival also when stratifying for molecular and clinical confounders on multivariate analysis including pre-operative tumor volume, MGMT promotor status, and non-surgical therapies (HR: 0.65, CI: 0.5-0.8; p = 0.001); and ≤1 cm 3 residual CE tumor translated into improved survival compared to non-surgical management. Accordingly, ‘maximal resection’ (class 2) had superior survival compared to ‘submaximal resection’ (class 3) (median OS after recurrence: 12 vs. 9 months; p = 0.003). Adjuvant chemotherapy further augmented the beneficial effects of lower residual CE tumor. Conversely, ‘supramaximal resection’ of non-CE tumor (class 1) was not associated with prolonged survival but frequently accompanied by post-operative deficits, hampering further treatment. The prognostic role of residual CE tumor was confirmed in propensity score analyses. Conclusions: Extent of resection for recurrent glioblastoma as quantified by residual CE tumor is highly prognostic and the RANO resect classification may serve to stratify patients accordingly. Chemotherapy may favorably contribute to the prognostic associations of re-resection. When pursuing resection of non-CE tumor, intraoperative mapping strategies to minimize the risk of post-operative deficits are recommended.