Abstract 4306: A prognostic model for clinical response to bevacizumab in recurrent glioblastoma multiforme

Abstract

Abstract Background: Recurrent glioblastoma multiforme (GBM) is an aggressive and deadly disease with limited treatment options. In responding patients bevacizumab-containing therapy prolongs survival and improves quality of life, although the beneficial effect varies greatly. The impact of prognostic factors in recurrent GBM patients has not been studied in detail and results are inconsistent. More importantly, no validated predictive baseline markers associated with a clinical response to bevacizumab therapy have been identified. Furthermore, GBM is a very heterogenic tumor and has been shown to change molecular pattern with time and as a result of treatment. The primary goal of this study was to describe molecular characteristics in GBM tumors at initial diagnosis and at recurrence and relate these data, including clinical baseline factors, to clinical outcome with the aim of identifying predictive factors for clinical bevacizumab response. Based on these data we will also elucidate to what extent molecular markers change expression from initial GBM diagnosis to time of relapse. Materials and Methods: For generation of prognostic models 219 recurrent GBM patients treated with bevacizumab plus irinotecan according to a previous published clinical protocol were included. For biomarker analysis, routine molecular analyses were available for 147 of the included patients at time of initial diagnosis and 81 patients at time of relapse prior to bevacizumab therapy. The candidate biomarkers constituted an immunohistochemistry (IHC) panel of EGFR, P53, MGMT and IDH1 and polymerase chain reaction analysis of 1p19q-status. Tumor samples from initial diagnosis and from time of relapse were analyzed by the χ2 test and the Wilcoxon test in order to evaluate changes in IHC expression patterns. Multiple candidate prognostic factors were screened by univariate logistic regression and Cox regression analysis modeling response and survival endpoints and variables with P-values of less than 0.10 were considered for multivariate analysis. Factors with a P-value of less than 0.05 in multivariate analysis were included in the prognostic models for response, progression-free survival (PFS) and overall survival (OS). Results: In multivariate analysis, corticosteroid use had a negative predictive impact on response at first evaluation (OR = 0.45; 95% CI: 0.22-0.93; P = 0.030) and at best response (OR = 0.51; 95% CI: 0.26-1.02; P = 0.056). Independent prognostic factors (P<0.05) negatively associated with PFS and OS were corticosteroid use, neurocognitive deficit and multifocal disease. Based on the three identified factors a prognostic model for overall survival at 6 and 12 months was developed. Significant prognostic and potentially predictive molecular biomarkers will be added to the models and results will be presented. Citation Format: Thomas Urup, Signe Regner Michaelsen, Camilla Bjørnbak Holst, Anders Toft, Ib Jarle Christensen, Kirsten Grunnet, Michael Kosteljanetz, Helle Broholm, Ulrik Lassen, Hans Skovgaard Poulsen. A prognostic model for clinical response to bevacizumab in recurrent glioblastoma multiforme. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4306. doi:10.1158/1538-7445.AM2015-4306