Systemic Inflammatory Markers for Prediction of Bevacizumab Benefit in Glioblastoma Multiforme.

Objective To evaluate the predictive significance of systemic inflammation markers (SIMs) in patients with glioblastoma multiforme (GBM), who were treated with bevacizumab (Beva). Study design Descriptive study. Place and duration of study The study was conducted at the Bezmialem Vakif University School of Medicine Hospital, Istanbul, Turkey, from January 2014 to September 2019. Methodology A total of 107 patients, 49 (45.8%) female and 58 (54.2%) male, were retrospectively included in the study. The cut-off values for the SIMs-C-reactive protein to albumin ratio (CAR), neutrophil to lymphocyte (NLR) platelet to lymphocyte ratio (PLR), and systemic immune-inflammatory index (SIII))-were defined by receiver operating characteristic (ROC) analysis. Overall survival (OS) was plotted using the Kaplan-Meier method and compared using the log-rank test. Cox regression analysis was performed for univariate and multivariate analyses. Results ROC analysis was performed to determine the optimal prognostic value of each parameter. CAR: 1.32, NLR: 2.9, PLR: 159, and SIII: 785 were determined as cut-off values for predicting OS based on the areas under the curve (AUC) in the ROC analysis. CAR at 0.626, had sensitivity of 67%, and specificity of 71% (p=0.129); NLR at 0.725 had sensitivity of 67%, and specificity of 79% (p=0.007); PLR at 0.675 had sensitivity of 67%, and specificity of 64% (p=0.036); and SIII at 0.685, had sensitivity of 56%, and specificity of 71% (p=0.026). A multivariate analysis demonstrated that CAR (p=0.006) and PLR (p=0.024) were independent prognostic factors for OS in patients with GBM, treated by Beva. Conclusion The present study's findings suggest that pretreatment CAR and PLR might be an independent predictive marker for patients with GBM, who are treated by Beva. Key Words: C-reactive protein-to-albumin ratio (AR), Glioblastoma multiforme, Neutrophil-to-lymphocyteratio (NLR), Platelet-to-lymphocyte ratio (PLR), Predictive score.