MAGMA: Multi-arm glioblastoma Australasia phase 3 trial.

2004 Background: MAGMA (ACTRN12620000048987; COGNO cooperative trials group) is a prospective, open-label, partial factorial randomized platform trial with optional randomization, aiming to improve treatments for glioblastoma (GBM). Two initial hypotheses were tested: Question 1 (Q1), ‘early’ daily temozolomide (TMZ) as soon as possible postop, pre-radiotherapy (RT); and Question 2 (Q2), ‘extended’ adjuvant TMZ beyond 6 cycles. Prior phase 2 studies suggest longer survival with early TMZ and inconsistent outcomes with extended adjuvant TMZ. MAGMA used baseline randomization with molecular stratification aiming for robust randomized evidence. Methods: Participants (pts) with newly diagnosed GBM (CNS WHO 2016) could be randomized to each question, stratified by extent of resection and IDH status (Q1 and Q2) and MGMT methylation (Q2 only). For Q1, pts were randomized to daily TMZ (75 mg/m 2 ) initiated as soon as possible after surgery followed by standard RT-TMZ (either 15 or 30 fractions), or RT-TMZ alone. For Q2, pts were randomized prior to adjuvant TMZ, to stop after 6 cycles or continue until progression. Primary outcome was overall survival (OS) from initial surgery. Secondary outcomes included progression-free survival (PFS), time to next treatment, OS from each randomization, clinically significant toxicity and health-related quality of life. Results: From September 2020 to July 2023, 347 pts consented across 28 Australian sites including 6 regional sites. Median follow-up was 40 months; final data cutoff was 31 July 2025. Mean age was 60 years (range 21-85), 60% were male, 92% IDH-wildtype, 42% MGMT-methylated, 56% had gross total resection and 13% biopsy-only. For Q1 (n=259), pts randomized to early TMZ (n=131) started TMZ at median 21 days post-surgery with mean exposure 13 days. Concurrent TMZ exposure, time to starting RT (median 34 vs 33 days), RT completion and toxicity (G3+ 34% vs 31%) were similar between arms. Median OS (months; 95% CI) was 16 (14-20) with early TMZ vs 20 (17-22) for control, HR 1.18 (0.89-1.56), p=0.24. Median PFS was 8 (7-9) vs 9 (7-12), HR 1.15 (0.88-1.52), p=0.31. For Q2 (n=291), among pts randomized to extended adjuvant TMZ (n=146), only 42% remained progression-free through 6 cycles and received more than 6 adjuvant TMZ cycles. Toxicity was similar between arms (G3+ 34% vs 31%). Median OS was 20 months (17-22) with extended adjuvant TMZ vs 18 (15-20) for control, HR 0.82 (0.63-1.06), p=0.13; after multivariable analysis, HR 0.73 (0.56-0.95). Median PFS was 9 (8-9) vs 9 (7-10), HR 0.88 (0.69-1.13), p=0.32. Conclusions: Early TMZ does not improve OS in pts with newly diagnosed GBM. Extended adjuvant TMZ beyond 6 cycles was safe, with insufficient evidence of improved OS in the primary analysis, although moderate effects remain plausible. Notably, many pts had progression before completing 6 cycles. Collaboration enables large-scale, pragmatic cancer trials across diverse healthcare settings in Australia. Clinical trial information: ACTRN12620000048987.