GOLD
Ultra-hypofractionated versus conventional chemoradiation for newly diagnosed glioblastoma: Survival and toxicity results of a multicenter randomized trial
Clinical Question
In patients with newly diagnosed glioblastoma undergoing temozolomide chemoradiation, is an ultra-hypofractionated 6x6 Gy regimen over 2 weeks non-inferior to standard 30x2 Gy over 6 weeks with respect to 2-year overall survival?
Bottom Line
Ultra-hypofractionated chemoradiation (6x6 Gy in 2 weeks) with temozolomide failed to demonstrate non-inferiority versus standard 30x2 Gy in newly diagnosed glioblastoma and was associated with markedly worse overall and progression-free survival after 6 months (median OS 13.0 vs 21.0 months) and a >5-fold higher risk of radiation necrosis/pseudoprogression (47.8% vs 16.2%). The GOLD trial provides randomized evidence against replacing standard chemoradiation with this accelerated regimen.
Major Points
- Open-label, multicenter, randomized phase III non-inferiority trial in 8 Dutch radiotherapy centers (Nov 5, 2020 - Mar 25, 2024); planned N=474 but stopped early at N=135 (67 experimental, 68 control) after funding was discontinued for slow accrual.
- 1:1 randomization stratified by MGMT methylation status, age (≤70 vs >70), prior surgery (biopsy vs resection) and center; primary endpoint 2-year overall survival with non-inferiority margin HR 1.20 (one-sided α=0.05, 80% power).
- Experimental arm: 6x6 Gy to CTV36Gy (GTV+5 mm) + 6x4 Gy to CTV24Gy (GTV+15 mm) over 2 weeks + concurrent and up to 6 cycles adjuvant temozolomide. Control: 30x2 Gy to CTV (GTV+15 mm) over 6 weeks + same temozolomide.
- Primary outcome: median OS 13.0 months (95% CI 10.3-15.7) with 6x6 Gy vs 21.0 months (95% CI not estimable) with 30x2 Gy; log-rank p<0.001. Cox proportional hazards violated → time-split HR 0.99 (0.39-2.50) at 0-6 mo, HR 2.54 (1.57-4.09), p<0.001 after 6 mo. 1-year survival 52.2% vs 70.6%.
- Progression-free survival: median 9.0 mo (6.8-11.2) vs 10.0 mo (6.8-13.2); log-rank p=0.060; time-split HR 2.08 (1.28-3.38), p=0.003 after 6 mo.
- Toxicity: no grade 4-5 events; radiation necrosis/pseudoprogression 32/67 (47.8%) vs 11/68 (16.2%); HR 5.20 (95% CI 2.56-10.58, p<0.001); median time to necrosis 13.0 months vs not reached; 1-yr cumulative probability 48% vs 19%.
- Rescue treatment for radiation-induced edema/necrosis: bevacizumab in 21/62 (34%) vs 5/61 (8%), p<0.001; dexamethasone use at 6-12 months 60-63% vs 31-32%, p≤0.006.
- Median cumulative concurrent temozolomide dose was lower with 6x6 Gy (900 vs 3000 mg/m²) due to shorter concurrent phase; adjuvant temozolomide completion similar (65% vs 70%).
- Exploratory subgroup analysis suggested the unfavorable effect was greater in patients ≤70 years (HR 2.20, 95% CI 1.41-3.43) and MGMT-methylated tumors (HR 2.99, 95% CI 0.95-9.44).
- Conclusion: non-inferiority of the 6x6 Gy regimen was not demonstrated; the regimen was associated with inferior survival and higher radiation necrosis and should not replace standard 30x2 Gy chemoradiation in glioblastoma.
Design
Study Type: Open-label, multicenter, randomized phase III non-inferiority trial
Randomization: 1
Blinding: Open-label (unblinded); computer-based randomization tool ensured allocation concealment until enrollment
Enrollment Period: November 5, 2020 - March 25, 2024
Follow-up Duration: Median 24 months (administratively censored at 24 months)
Centers: 8
Countries: Netherlands
Sample Size: 135
Power Calculation: Assumed 2-year mortality 73% (control) and 66% (experimental); anticipated HR 0.90, one-sided α=0.05, 80% power, non-inferiority margin HR 1.20 → 299 deaths / 430 patients required; planned N=474 with 10% attrition. Enrollment stopped early at 135 after funding discontinuation.
Analysis: Intention-to-treat primary analysis (per-protocol as sensitivity); Kaplan-Meier estimates; one-sided log-rank non-inferiority test; Cox proportional hazards with 95% CIs. Because hazards were non-proportional, a post-hoc time-dependent Cox model with a 6-month split was used descriptively. Fisher's exact test for toxicity; SPSS v31.0.
Inclusion Criteria
- Adult (≥18 years)
- Newly diagnosed histologically confirmed glioblastoma per WHO 2016 criteria
- Prior biopsy or resection
- Karnofsky Performance Status ≥70
- Eligible for standard temozolomide chemoradiation
- (Full eligibility criteria provided in supplementary material S1)
Exclusion Criteria
- (Full exclusion criteria listed in supplementary material S1; not enumerated in the main paper)
Arms
| Field | Ultra-hypofractionated (6x6 Gy) chemoradiation | Control |
|---|---|---|
| Intervention | 6 fractions of 6 Gy to CTV36Gy (GTV+5 mm) plus 6 fractions of 4 Gy to CTV24Gy (GTV+15 mm), 3 fractions/week over 2 weeks (simultaneous integrated boost-like schedule) with IMRT/VMAT, concurrent daily temozolomide for 2 weeks and up to 6 cycles adjuvant temozolomide. | 30 fractions of 2 Gy to CTV (GTV+15 mm), 5 fractions/week over 6 weeks with IMRT/VMAT, concurrent daily temozolomide for 6 weeks and up to 6 cycles adjuvant temozolomide. |
| N | 67 | 68 |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Overall survival at 2 years, defined as time from first day of radiotherapy (or planned RT start date for patients who did not start RT) to death from any cause; primary non-inferiority test with margin HR 1.20 (one-sided α=0.05). | Primary | 36/68 (52.9%) deaths; median OS 21.0 months (95% CI not estimable); 1-year survival 70.6% | 56/67 (83.6%) deaths; median OS 13.0 months (95% CI 10.3-15.7); 1-year survival 52.2% | ||
| Progression-free survival (from first day RT to MRI-confirmed recurrence or death) | Secondary | 51/68 (75.0%) events; median 10.0 months (95% CI 6.8-13.2) | 60/67 (89.6%) events; median 9.0 months (95% CI 6.8-11.2) | ||
| Any bevacizumab administration during follow-up | Secondary | 5/61 (8%) | 21/62 (34%) | <0.001 | |
| Dexamethasone use at 6-12 months post-RT | Secondary | 31-32% | 60-63% | ≤0.006 | |
| Concurrent temozolomide completion | Secondary | 58/64 (91%) | 65/65 (100%) | ||
| Adjuvant temozolomide (6 cycles) completion | Secondary | 45/64 (70%) | 42/65 (65%) | ||
| Second-line treatment after documented progression | Secondary | 18/50 (36.0%) | 24/54 (44.4%) | ||
| Radiation necrosis or pseudoprogression (reported clinically) | Safety | 11/68 (16.2%); median time not reached; 1-yr cumulative 19% (6.7-31.3%) | 32/67 (47.8%); median time to occurrence 13.0 months (95% CI 9.0-17.0); 1-yr cumulative 48% (33.1-62.9%) | 5.2 | <0.001 |
| Grade 4-5 toxicities (CTCAE v5.0) | Safety | None reported | None reported | ||
| Radiotherapy interruptions | Safety | 2 temporary interruptions (hospitalization, headache); all completed assigned RT | 0 (all completed assigned RT) |
Subgroup Analysis
Exploratory subgroup Cox analyses (unadjusted for multiple comparisons) suggested a more pronounced unfavorable effect of 6x6 Gy in patients ≤70 years (HR 2.20, 95% CI 1.41-3.43; 83/125 events) vs >70 years (HR 0.97, 95% CI 0.25-3.72; 9/10 events), in MGMT-methylated tumors (HR 2.99, 95% CI 0.95-9.44) and MGMT-unknown (HR 3.50, 95% CI 1.72-7.12) vs MGMT-unmethylated (HR 1.29, 95% CI 0.69-2.38), and in patients who underwent resection (HR 1.95, 95% CI 1.25-3.04) vs biopsy (HR 2.92, 95% CI 0.74-11.56). Overall HR 2.06 (95% CI 1.35-3.14). Multivariable Cox adjusting for age, KPS, MGMT and extent of resection confirmed unfavorable outcomes in the ultra-hypofractionated group.
Criticisms
- Early termination at 135/474 planned patients (funding discontinued for slow accrual) - the trial was substantially underpowered for the primary non-inferiority endpoint.
- The proportional-hazards assumption was violated, precluding meaningful interpretation of the prespecified HR margin of 1.20 over the full follow-up; time-split analyses were post-hoc and descriptive only.
- Sample-size calculation used an anticipated HR of 0.90 favoring the experimental arm, which reduced the planned N compared with an assumption of equal efficacy; historical event rates from clinically different populations introduce constancy-assumption uncertainty.
- Open-label design (blinding not feasible for radiotherapy fractionation).
- No central review of radiotherapy plans, allowing variability across the 8 centers.
- MRI frequency and use of advanced imaging (e.g., PET) were not mandated - misclassification between tumor progression, pseudoprogression and radiation necrosis cannot be excluded; central imaging review is ongoing.
- Median cumulative concurrent temozolomide dose was much lower in the experimental arm (900 vs 3000 mg/m²) due to shorter concurrent phase, potentially contributing to survival difference; unlikely to fully explain magnitude and time-dependent pattern.
- The experimental regimen did not include bevacizumab (given with the same 6x6 Gy schedule in Omuro 2014), which may have mitigated the observed radiation-necrosis excess - direct comparison to prior phase II data is limited.
- Subgroup analyses were exploratory and not adjusted for multiplicity/interaction.
- MRI imaging schedule differed by center and was not predefined in the protocol.
Funding
ZonMw, the Netherlands Organization for Health Research and Development (project number 852002028). ZonMw had no role in study design, data collection, analysis, interpretation, or writing.
Based on: GOLD (Radiotherapy and Oncology, 2026)
Authors: de Jong AM, van der Boog ATJ, Wester G, et al; GOLD Trial Investigators
Citation: Radiother Oncol 2026;223:111726. DOI: 10.1016/j.radonc.2026.111726
Content summarized and formatted by NeuroTrials.ai.