Background: A meta-analysis of chemotherapy for glioblastoma multiforme (GBM) was performed. We sought to update prior analyses by focusing exclusively on GBM, including new trials of novel treatments, assessing effectiveness of individual treatment categories and presenting data in a clinically useful format.
Methods: A search of MEDLINE and EMBASE was conducted for randomised controlled trials of chemotherapy in GBM.
Results: Relative risks (RRs) for survival in 16 trials comparing chemotherapy with no chemotherapy were 1.18 (95% CI 1.08, 1.30) at 6 months, 1.53 (95% CI 1.26, 1.86) at 12 months and 2.12 (95% CI 1.60, 2.80) at 24 months. Nitrosourea compounds, local therapy (e.g. carmustine [1,3-bis [2-chloroethyl]-1-nitrosourea] wafers) and temozolomide were all more effective than no chemotherapy. Absolute increases in survival at 6, 12 and 24 months were 11%, 8% and 1%, respectively, for nitrosourea compounds; 8%, 24% and 5%, respectively, for local therapy; and 4%, 15% and 17%, respectively, for temozolomide. Efficacy of local therapy and temozolomide peaked at 12 and 18 months, respectively. After 2 years, nitrosourea compounds no longer provided clinically relevant benefit (number needed-to-treat [NNT] = 100; effect size [ES] = 0.17 SD), local therapy had diminishing returns (NNT = 20) that remained clinically relevant (ES = 0.71 SD) and temozolomide continued to show good efficacy (NNT = 5.9; ES = 0.74 SD). Survival was not significantly improved with multi-agent versus single-agent nitrosourea-based therapy in five trials: 6-month RR 0.91 (95% CI 0.71, 1.16); 24-month RR 1.33 (95% CI 0.72, 2.46).
Conclusion: Although nitrosourea compounds, local therapy and temozolomide are all effective in the treatment of GBM, local therapy and temozolomide may be associated with greater response, with clinically significant benefits extending to 24 months. The timing of peak benefits of local and temozolomide therapy suggests this combination may be more effective than single-agent chemotherapy and warrants further study.