Abstract PURPOSE The goal of our project was to define the molecular features of recurrent glioblastoma (rGBM) and their prognostic significance. METHODS We retrospectively reviewed 121 consecutive glioblastoma patients who underwent re-resection at a single institution from 2016-2021. Primary treatment in all patients included maximal safe resection followed by radiotherapy with concurrent and adjuvant temozolomide (TMZ). Capture-based targeted DNA sequencing for point mutations and copy number analysis was performed on all re-resection samples. Overall (OS) and progression-free survival (PFS) from re-resection were analyzed with the Kaplan-Meier method and Cox regression analyses. RESULTS Median age at first recurrence was 57 years (range, 27-78) and follow up from re-resection was 11 months. Salvage therapy comprised re-irradiation (n = 52, 43%), temozolomide (TMZ) (n = 39, 32%), lomustine (n = 47, 39%), and/or bevacizumab (n = 22, 18%). Median OS and PFS were 12.1 and 4.4 months, respectively. Median tumor mutation burden (TMB) was 4 mut/Mb (range, 1-650), average unstable microsatellite sites (MSI) were 1.16% (range, 0-20), and MGMT was methylated in 71 (59%) patients. Longer OS was found after postoperative TMZ (19.1 vs. 9.1), KPS > 70 (13.6 vs. 11.1), and/or re-irradiation (12.6 vs. 8.4) (p< .05). On multivariate analysis, TMZ (HR 0.2, 0.079-0.48), re-irradiation (HR 0.27, 0.12-0.58), and mutations in PDGFRA (HR 0.24, 0.067-0.83) or PTEN (HR 2.3, 1.1-4.8) genes were associated with longer OS after re-resection (p< .05). Multivariate predictors of re-irradiation response included MSI >1.16 (HR 0.25, 0.084-0.75), and mutations in PDGFRA (HR 0.04, 0.005-0.33), TERT (HR 0.16, 0.033-0.78), or PIK3R1 (HR 0.18, 0.041-0.8) genes. In the 46 matched primary/recurrent tumors, 40 (87%%) had a change in a mutation and 9 (20%) demonstrated hypermutation, of which 4 (44%) had mismatch repair deficiencies. CONCLUSIONS Our study outlines the molecular factors associated with survival and response to re-irradiation in rGBM as well as, the molecular differences between primary/recurrent samples.
Read full abstract