INTRAGO-II
Dose escalation with intraoperative radiotherapy in newly diagnosed glioblastoma (INTRAGO-II): an open-label, multicentre, randomised, controlled, phase 3 trial
Bottom Line
A single 30 Gy intraoperative radiotherapy boost added to standard chemoradiotherapy did NOT improve PFS (11.0 vs 11.4 months; HR 1.1, p=0.47) or overall survival, and did not alter local recurrence patterns, arguing against further local dose intensification in resectable glioblastoma.
Major Points
- Phase 3 international RCT (18 centres, 7 countries) of 30 Gy kilovoltage IORT added to standard chemoradiotherapy in newly diagnosed, resectable glioblastoma; 314 randomised, 298 in full-analysis set.
- Primary endpoint (masked, centralised, RANO-based PFS): 11.0 vs 11.4 months; HR 1.1 (95% CI 0.85–1.44); p=0.47 — clearly negative.
- Secondary endpoints also negative: median OS 17.7 vs 18.7 months (HR 1.2, 0.91–1.54; p=0.20); local recurrence within 1 cm of cavity 72% vs 71% (p=0.87).
- Subgroup analyses (age, sex, KPS, MGMT methylation, IDH1 status, residual disease) showed no benefit from IORT in any prespecified subgroup.
- Toxicity signal: numerically more grade 3–4 seizures (13% vs 7%), radiation necrosis (7% vs 2%, p=0.06), thromboembolism, and more serious AEs (65% vs 53%) and grade-5 events (14 vs 6) with IORT.
- Only 71% of IORT patients received the planned 30 Gy (29% received 20–25 Gy due to organ-at-risk constraints); most recurrences remained within the high-dose zone, arguing biology, not geometry, drives failure.
Design
Study Type: Randomised Controlled Trial (Phase 3)
Randomization: 1
Blinding: Open-label (participants, physicians, investigators unmasked; centralised imaging review masked to allocation)
Enrollment Period: Dec 9, 2016 – Jun 17, 2024
Follow-up Duration: Median 17.2 months (IQR 10.5–27.1)
Centers: 18
Countries: Germany, USA, Spain, Canada, China, South Korea, Brazil
Sample Size: 298
Analysis: Full-analysis set (all randomised except post-hoc ineligibility for non-glioblastoma histology [n=12] or informed-consent deficiencies [n=4]); intention-to-treat by allocated group
Inclusion Criteria
- Age 18–80 years
- Newly diagnosed supratentorial glioblastoma amenable to resection of contrast-enhancing tumour
- Karnofsky performance score ≥60%
- Intraoperative feasibility of intraoperative radiotherapy (adequate cavity and haemostasis after maximal safe resection)
- Adequate organ function
- Written informed consent before surgery
Exclusion Criteria
- Multicentric disease
- Previous cranial radiotherapy
- Recent cytotoxic therapy
- Substantial competing comorbidity limiting life expectancy
- Previous antiangiogenic therapy
- MRI contraindications
- Pregnancy or breastfeeding
- Estimated intraoperative radiotherapy dose to adjacent organs at risk exceeding 8 Gy
Arms
| Field | Intraoperative radiotherapy | Control |
|---|---|---|
| Intervention | Maximal safe resection + single 30 Gy kilovoltage IORT to resection cavity (INTRABEAM system, dose reductions to 20 Gy allowed for organ-at-risk constraints) + standard postoperative EBRT 60 Gy with concurrent temozolomide 75 mg/m² followed by six 28-day cycles of adjuvant temozolomide 150–200 mg/m² (days 1–5) | Maximal safe resection alone + standard postoperative EBRT 60 Gy with concurrent and adjuvant temozolomide (as above); lomustine or tumour-treating fields permitted per amendment |
| Duration | Postoperative treatment initiated 3–5 weeks after surgery; adjuvant chemo starts 4 weeks after EBRT | Same timing as active arm |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Median progression-free survival (time from randomisation to radiological progression per RANO with masked central review, new/salvage therapy, unequivocal clinical deterioration, or death from any cause) | Primary | 11.4 months (95% CI 9.7–13.9) | 11.0 months (95% CI 9.2–12.6) | 1.1 | 0.47 |
| Median overall survival | Secondary | 18.7 months (17.2–21.1) | 17.7 months (15.5–21.1) | 1.2 | 0.20 |
| Local recurrence within 1 cm of cavity (pattern of failure, central review) | Secondary | 71% (60/85) | 72% (76/106) | 0.87 | |
| Deaths during study | Secondary | 104 (98 not specified, 6 AE-related) | 129 (116 not specified, 13 AE-related) | N/R | |
| Study discontinuation before endpoint | Secondary | 20% (28/137) | 12% (19/161) | N/R | |
| Serious adverse events (any) | Adverse | 65% (105/161) IORT vs 53% (72/137) SOC (252 vs 142 events) | |||
| SAEs at least possibly related to treatment | Adverse | 38% of patients (65 events) IORT vs 26% (27 events) SOC | |||
| Grade 5 AEs attributable to CTCAE terms | Adverse | 14 (IORT: CNS toxicity x2, MI x2, sepsis x2, cardiac arrest, fever, lung infection, fracture, postoperative haemorrhage, neoplasm, haematoma, unspecified death) vs 6 (SOC: lung infection x2, multiorgan failure, encephalitis, neoplasm, cystitis) | |||
| Grade 3–4 seizure | Adverse | 13% (21) vs 7% (9); p=0.064 | |||
| Grade 3–4 radiation necrosis | Adverse | 7% (11) vs 2% (3); p=0.06 | |||
| Grade 3–4 muscle weakness | Adverse | 7% (12) vs 14% (19) | |||
| Grade 3–4 thrombocytopenia | Adverse | 4% (7) vs 8% (11) | |||
| Any-grade fatigue | Adverse | 43% (69) vs 34% (46) | |||
| Any-grade headache | Adverse | 42% (67) vs 34% (47) | |||
| Any-grade seizure (grade 1–2) | Adverse | 20% (32) vs 16% (22) | |||
| Wound complications | Adverse | 5% (8) vs 1% (2) | |||
| Wound infections | Adverse | 4% (6) vs 3% (4) | |||
| Intracranial haemorrhage grade 3–4 | Adverse | 3% (5) vs 1% (1) | |||
| Thromboembolic event grade 3–4 | Adverse | 4% (7) vs 3% (5); grade 5: 2% (3) vs 0 | |||
Subgroup Analysis
Prespecified subgroups (sex, age, KPS, chemotherapy administration, MGMT methylation, IDH1 mutation, residual disease) showed no significant difference in PFS or OS with IORT vs SOC; effect null across all clinical subgroups.
Criticisms
- Open-label design likely drove differential attrition (20% control vs 12% IORT discontinuation), potentially inflating events in the SOC arm.
- Only 71% of IORT patients received the planned 30 Gy; 29% received reduced 20–25 Gy for organ-at-risk sparing, diluting the intended dose-escalation signal.
- Cavity-directed boost cannot address microscopic disease infiltrating several centimetres beyond the contrast-enhancing margin — an inherent geometric limitation of any local IORT approach.
- Modest sample size (298 in full-analysis set) with heterogeneous molecular subgroups; MGMT status was determined locally with no central pathology review.
- Substantial portion of enrolment overlapped the COVID-19 pandemic, disrupting treatment delivery at multiple centres.
- Protocol amendments during the trial (allowing lomustine per CeTeG and tumour-treating fields) introduced variability in the standard-of-care backbone.
- Included patients with KPS as low as 60%, a higher-risk population than most glioblastoma registration trials — may limit direct comparability.
Funding
Universities of Heidelberg and Bonn (sponsors); Carl Zeiss Meditec (research funds and device manufacturer); Deutsche Forschungsgemeinschaft (SFB 1389 UNITE Glioblastoma); German Federal Ministry of Research, Technology and Space (BMFTR).
Based on: INTRAGO-II (Lancet Oncology, 2026)
Authors: Giordano FA, Ganslandt O, Münter MW, et al.
Citation: Lancet Oncol. 2026;27(7):864-878. DOI: 10.1016/S1470-2045(26)00235-4
Content summarized and formatted by NeuroTrials.ai.