EPID-13. IDENTIFICATION OF PROGNOSTIC MARKERS IN A COHORT OF CONSECUTIVE NON-SELECTED GLIOBLASTOMA PATIENTS RECEIVING STANDARD THERAPY

Abstract

Abstract Glioblastoma is treated by standard with maximal surgical resection followed by radiation therapy with concomitant and adjuvant temozolomide. Still, median overall survival (OS) for glioblastoma is around 15 months, although the prognosis varies greatly. To improve the treatment, stratification of glioblastoma patients according to favorable and poor prognostic groups is needed. The aim of this study was to develop a prognostic model for glioblastoma patients treated with standard therapy outside clinical trials. The study included 680 consecutive, non-selected glioblastoma patients with WHO performance status (PS) 0–2 administered postoperative with standard first-line therapy between years 2005–2016 at the University Hospital of Copenhagen, Denmark. At disease recurrence, patients were mostly treated by reoperation and/or bevacizumab combination therapy. Patients were retrospectively evaluated by reviewing of medical records, histopathological descriptions and imaging analyses. The cohort had a median progression-free survival (PFS) of 7.5 months, median OS of 15.7 months and median OS from recurrence of 7.4 months. Comparative analyses of survival endpoints dependent on year of diagnosis did not reveal any statistically significant improvement from 2005 and forth. Factors significantly associated with worse prognosis in univariate and multivariate analyses were for OS, low extent of primary surgery, increased age, multifocal disease, increased PS and O6-methylguanine DNA methyltransferase (MGMT) tumor methylation. For PFS, multifocal disease, increased PS, and MGMT tumor methylation were significantly associated with worse prognosis. Gender, tumor location, tumor size, comorbidity score and site of tumor relapse were not significantly associated with any survival endpoints. Based on study results, a predictive model for survival will be generated able to calculate survival probability for the individual glioblastoma patient. Such a model could prove useful for stratification in clinical scientific protocols and for making glioblastoma treatment more individualized. Updated results will be presented.