Comparative efficacy of therapeutic strategies for recurrent or refractory glioblastoma: a systematic review and network meta-analysis.

Recurrent or refractory glioblastoma has a poor prognosis, and no universally accepted standard treatment has been established. Randomized evidence is fragmented, making pairwise comparisons insufficient. This study aimed to compare available treatments using a frequentist network meta-analysis. PubMed, Web of Science, and the Cochrane Library were searched from inception to April 2026 for randomized controlled trials involving adults with recurrent or refractory glioblastoma. Prespecified outcomes were progression-free survival (PFS), overall survival (OS), and objective response rate (ORR). Hazard ratios (HRs) were used for PFS and OS, and odds ratios (ORs) for ORR. Random-effects frequentist network meta-analyses were conducted, with P-scores used as exploratory rankings. Sensitivity analyses excluded trials with fewer than 20 patients in any treatment arm. The protocol was registered with PROSPERO (CRD420261398465). Twenty-four studies were included in the qualitative synthesis, of which 19 contributed to at least one network meta-analysis. Compared with lomustine, bevacizumab plus lomustine significantly improved PFS (HR = 0.57, 95% CI: 0.40-0.80), and remained significant after excluding small trials (HR = 0.58, 95% CI: 0.34-0.97). Bevacizumab plus vorinostat had the highest numerical P-score for PFS, although its effect was not statistically significant. No intervention showed a statistically significant OS advantage over bevacizumab. Rindopepimut plus bevacizumab had the most favorable numerical OS estimate (HR = 0.53, 95% CI: 0.23-1.21), but this was non-significant and derived from an EGFRvIII-positive population. No treatment significantly improved ORR versus bevacizumab, whereas nivolumab was associated with lower odds of response (OR = 0.28, 95% CI: 0.14-0.57). Excluding small trials had limited influence on PFS and OS findings but altered the ORR ranking by removing the highly ranked ERC1671-based regimen. Bevacizumab plus lomustine showed the clearest PFS benefit, but the absence of an OS benefit precludes interpretation as broadly superior. No treatment showed a confirmed OS advantage, and the favorable numerical signal for rindopepimut plus bevacizumab should not be generalized beyond EGFRvIII-positive patients. P-score rankings should be considered exploratory and interpreted alongside effect estimates, confidence intervals, sample sizes, molecular eligibility, safety, and treatment burden. https://www.crd.york.ac.uk/PROSPERO/view/CRD420261398465, Identifier: CRD420261398465.