Which treatment is best? A systematic review and Bayesian network meta-analysis of adjuvant treatments for low-grade glioma.

Abstract

2057 Background: The standard of care for low-grade glioma (LGG) includes resection with or without adjuvant treatment. It is currently unknown which adjuvant strategy is best. Methods: A systematic review of PubMed and EMBASE for randomized controlled trials (RCTs) evaluating adjuvant treatment for LGG published on or before 6/2/2025 was conducted as pre-registered (CRD420251072899). Study eligibility and data extraction were reviewed by two authors independently, and risk of bias was assessed via Cochrane RoB2. Two Bayesian network meta-analyses profiling de novo LGG were conducted with uninformative priors adjusting for baseline characteristics of each trial cohort. The pre-specified outcome measure was the distribution of posterior HR PFS (primary meta-analysis) and HR OS (secondary meta-analysis) for each identified treatment against observation. Posterior distributions were summarized with the posterior median hazard ratio (HR), and 95% equal-tailed credible interval (CrI). Further the probability of major benefit (HR≤0.6) and any magnitude harm (HR>1.0) were reported for each meta-analysis. Results: Six RCTs (all with low risk of bias) profiling 1,757 participants were identified for the primary meta-analysis (PFS). Seven RCTs (6 with low risk of bias 1 with high risk of bias) profiling 1,392 patients were identified for the secondary meta-analysis (OS). In total, six adjuvant treatment strategies were identified: radiotherapy (RT) + procarbazine/CCNU/vincristine (PCV), RT + temozolomide (TMZ), RT + CCNU alone, high-dose RT (HDRT, >54Gy), low-dose RT (LDRT; ≤54Gy), and TMZ. The posterior median HRs and 95% CrIs are reported in Table 1. For all profiled treatments, the probability of any benefit (HR<1.0) on the endpoint of PFS was >98% and >76% for OS. Additionally, the probability of major benefit (P1) and any magnitude harm (P2) were reported as clinically significant summary measures of all posterior distributions (Table 1). Conclusions: Based on the identified RCTs, RT+PCV was found to have the most favorable probability efficacy and RT+TMZ a similar distribution of treatment effect. RT+PCV and RT+TMZ had a >98% probability of any benefit for both endpoints. The probability of major benefit for RT+PCV and RT+TMZ was >99.9% and 84.8% for PFS and 71.3% and 63.3% for OS. Probability of benefit/harm of adjuvant therapy for low-grade glioma. PFS P1 PFS P2 PFS OS P1 OS P2 OS PT+PCV 0.29 (0.19-0.43) >99.9% <0.1% 0.53 (0.32-0.84) 71.3% 0.4% RT+TMZ 0.45 (0.27-0.76) 84.8% 0.2% 0.55 (0.33-0.91) 63.3% 1.1% RT+CCNU - - 0.63 (0.28-1.42) 46.0% 13.4% HDRT 0.59 (0.41-0.83) 55.0% 0.2% 0.86 (0.57-1.30) 4.4% 23.0% LDRT 0.58 (0.44-0.76) 59.0% <0.1% 0.89 (0.63-1.24) 1.4% 23.4% TMZ 0.67 (0.47-0.97) 26.9% 1.4% - - - Observation 1 (Ref.) - - 1 (Ref.) - - P1, probability HR≤0.6; P2, probability HR>1.0.