The value of bevacizumab (BEV) in recurrent glioblastoma is unclear. Imaging parameters and progression-free survival (PFS) are problematic endpoints. Few data exist on clinical factors influencing overall survival (OS) in unselected patients with recurrent glioblastoma exposed to BEV. We retrospectively analyzed 174 patients with recurrent glioblastoma treated with BEV at two German brain tumor centers. We evaluated general patient characteristics, MGMT status, pretreatment, concomitant oncologic treatment and overall survival. Karnofsky performance score, number of prior chemotherapies, number of prior recurrences and combined treatment with irinotecan (IRI) were significantly associated with OS in univariate analysis. We did not find differences in OS related to sex, age, histology, MGMT status, prior surgical treatment or number of prior radiotherapies. Combined treatment with IRI and higher KPS both remained significantly associated with prolonged survival in multivariate analysis, but patients receiving IRI co-treatment had less advanced disease. Grouping into clinically relevant categories revealed an OS of 16.9 months from start of BEV in patients with first recurrence and KPS ≥ 80 % (n = 25). In contrast, in patients with second recurrence and KPS < 80 %, OS was 3.6 months (n = 27). Our observational data support an early use of BEV in patients with good performance status. The benefit of co-treatment with IRI in our cohort seems to be the result of biased patient recruitment.
Keywords: Bevacizumab; Irinotecan; Karnofsky performance score; Recurrent glioblastoma; Survival.