FIRESTORM
ForHigh-risk meningioma (WHO gr 2 post-STR, recurrent gr 2, any gr 3), postop RT
HR 0.40
95% CI 0.24-0.69, P = .001 (MVA); IPTW HR 0.45 (0.24-0.83)
TL;DR5-yr PFS 65.8% vs 38.8% with dose-escalated postop RT (BED ≥79.2 Gy) in high-risk meningioma; MVA HR 0.40.
The benefit survives the obvious confounders: excluding the 35 single-fraction SRS cases from SD-RT, DE-RT still gave 5-yr PFS 65.8% vs 41.7% (HR 0.56, 0.36-0.86), and photon-alone escalation matched carbon on PFS (68.3% vs 61.1%) with 0% grade ≥3 radionecrosis. That makes 66-70 Gy conventionally fractionated photons the practical escalation route, not particle referral.
In a subtotally resected WHO grade 2 or any grade 3 meningioma being planned for postoperative RT, this supports considering escalation beyond 60 Gy/30 fx rather than defaulting to it; it does not speak to gross-totally resected grade 1 disease.
Photon-alone escalation to 70 Gy/35 fx or 66 Gy/33 fx matched carbon on 5-yr PFS (68.3% vs 61.1%) with zero grade ≥3 radionecrosis, versus 9.1% in the carbon subgroup. Boost PTV margins were 0.5 to 2 mm off cavity and residual tumor. That makes escalation a standard photon department decision, not a particle referral.
14 details
Individual patient-level meta-analysis pooling 248 patients from 7 international institutions, all retrospective except one prospective trial. Median follow-up 67.1 months (range 2.13-178). PFS was the primary outcome, analyzed by Kaplan-Meier, Cox MVA, and IPTW propensity weighting.
High-risk meningioma per RTOG 0539: newly diagnosed WHO grade 2 after STR/biopsy, any recurrent grade 2, or any grade 3. 188 (75.8%) were grade 2, 103 (41.5%) recurrent, and 182 (75.2%) had Simpson grade 4/STR. Median age 60; 55 (22.2%) had prior RT.
DE-RT was defined by biologically effective dose ≥79.2 Gy (equivalent to 66 Gy in 33 fractions) or receipt of a carbon-ion boost; anything below that threshold counted as SD-RT. Photon DE-RT used a sequential or simultaneous integrated boost with a 0.5 to 2 mm PTV margin, versus 1-2 cm CTV margins in the SD-RT arm. Carbon DE-RT boosted 16 Gy/8 fx after 50 Gy/25 fx photons with a 6 mm CTV margin.
Whole-cohort 3- and 5-yr PFS were 62.8% and 45.0%. DE-RT improved 3-yr (86.4% vs 55.6%) and 5-yr PFS (65.8% vs 38.8%), P = .0022, holding on stratified Cox by grade (HR 0.40, 0.22-0.73) and after IPTW (3-yr 84.7% vs 55.8%). OS was not improved on MVA or IPTW.
| Subgroup | 3-yr PFS DE-RT vs SD-RT | 5-yr PFS DE-RT vs SD-RT | MVA HR (95% CI) | P |
|---|---|---|---|---|
| Simpson 1-3 | 87.5% vs 55.9% | 70.0% vs 40.0% | 0.31 (0.08-1.14) | .08 |
| Simpson 4-5 | 86.3% vs 55.4% | 63.3% vs 38.4% | 0.55 (0.36-0.84) | .006 |
| Cohort | Any-grade RN | Grade ≥3 RN |
|---|---|---|
| DE-RT overall | 20 of 59 (33.9%) | 3 of 59 (5.1%) |
| DE-RT mixed carbon/photon | 15 of 33 (45%) | 3 of 33 (9.1%) |
| DE-RT photon-alone | 5 of 26 (19.2%) | 0% |
| SD-RT | 25 of 189 (13.2%) | 6 of 189 (3.2%) |
Any-grade radionecrosis was 33.9% with DE-RT vs 13.2% with SD-RT (P = .001), but grade ≥3 RN did not differ (5.1% vs 3.2%, P = .47). All 3 grade ≥3 events in the DE-RT arm occurred in the mixed carbon-photon cohort; the photon-alone DE-RT cohort had none, and its any-grade RN rate was not significantly higher than SD-RT (P = .41). One grade 5 event occurred in each arm.
Published series anchored on RTOG 0539 report 5-yr PFS of roughly 40% to 60% with 60 Gy/30 fx, which is what the SD-RT arm reproduces (45.0% overall). The escalation signal draws on MARCIE (carbon boost) and the Zeng et al. photon series, both contributors to this pool, so the comparison is partly internal rather than independent.
DE-RT was delivered at only 2 of 7 institutions, one carbon-only and one photon-only, so treatment arm is nearly collinear with center, and unmeasured practice differences (DOTATATE PET planning, response assessment, supportive care) travel with it. Molecular classification was unavailable, and P values were not adjusted for multiple testing across the subgroup analyses.
The Simpson 1-3 subgroup showed absolute separation as large as Simpson 4-5 (5-yr PFS 70.0% vs 40.0%) without reaching significance (P = .08 on MVA), which the authors attribute to a smaller GTR subset and fewer events rather than an absent effect. Whether escalation belongs in fully resected disease is the open question, and the PFS-only benefit means the case rests on avoiding local progression and its neurologic morbidity, not on survival.
Retrospective IPD pooling, DE-RT confined to 2 centers, no randomization; IPTW cannot remove selection. Direction consistent across every sensitivity analysis, but prospective randomization still needed.
- Benefit of dose escalation after gross total resection
- Whether molecular subgroups predict DE-RT benefit
- Prospective randomized confirmation of the PFS signal
📚 Sources · 📄 1 paper
Abstract
The longer read
The question this pooling addresses is narrow but real: 60 Gy in 30 fractions has been the postoperative default for high-risk meningioma largely by convention, and the 5-yr PFS it delivers (roughly 40% to 60% across published series, 45.0% in this cohort) has never been good enough for anyone to defend the dose as optimized. FIRESTORM's contribution is that the escalation signal is large, an absolute improvement the authors put at 20% to 30% at 5 years, and that it does not evaporate under the sensitivity analyses that usually dissolve retrospective dose comparisons.
That robustness matters because the naive read of this cohort has an obvious failure mode. The SD-RT arm contains 35 stereotactic radiosurgery cases, most treated to 12 to 18 Gy in a single fraction, whose 5-yr PFS was 20.5%, and half of the single-fraction cases had prior RT. A recurrent, previously irradiated meningioma treated with SRS is a salvage population, and pooling it into a comparator arm manufactures a dose-response relationship out of case mix alone. The authors saw this and removed it in stages: excluding radiosurgery, the effect persisted (HR 0.56, 0.36-0.86); restricting the comparator to conventionally fractionated 59.4 Gy/33 fx or 60 Gy/30 fx, it persisted (HR 0.51, 0.28-0.93); additionally excluding prior-RT patients, it persisted and strengthened (HR 0.37, 0.19-0.73). An effect that survives progressive removal of its most plausible confounders is behaving differently from one driven by them.
What should still move confidence downward is structural rather than statistical. DE-RT was available at only 2 of the 7 contributing institutions, one using carbon exclusively and the other photons exclusively. Treatment assignment is therefore nearly collinear with institution, and no propensity model can adjust for a covariate that is essentially predicted by center. Everything that differs between a carbon-ion center and general practice, DOTATATE PET for target definition, the threshold for calling progression on surveillance MRI, follow-up imaging intervals, rides along with the exposure. The IPTW analysis balanced prior surgery, prior RT, extent of resection, grade, and recurrence status, and it should be read as evidence that the measured imbalances are not driving the result, not as evidence that unmeasured ones are absent.
The technique detail is where the practical read lives, and it cuts against the assumption that escalation requires particles. Mixed carbon-photon and photon-alone DE-RT produced similar PFS (5-yr 61.1% vs 68.3%, P = .34), while toxicity separated sharply: 15 of the 20 any-grade radionecrosis events and all 3 grade 3 or higher events in the DE-RT arm came from the carbon subgroup, and photon-alone DE-RT had no grade 3 or higher events and no significant excess of any-grade RN over standard dose (P = .41). If the dose is what matters and the particle is not, then 66 to 70 Gy in conventional fractionation with tight boost margins is deliverable in a standard photon department, which is a very different implementation problem than referral to a carbon facility.
The honest limit is that this is a PFS result. OS was not improved on MVA (HR 0.66, 0.33-1.34) or IPTW (HR 0.77, 0.41-1.45), and with grade 3 disease, recurrence, and age dominating the survival model, the sample may simply be too small to resolve a survival effect that longer follow-up would show. The argument for escalating is that local progression in a high-risk meningioma carries real neurologic morbidity and limited salvage, so a durable local-control gain is worth having on its own terms. That is defensible, but it is a judgment about the value of the endpoint rather than a demonstration that patients live longer, and it leaves the case for prospective randomization intact.