NRG-CC009
ForSCLC brain mets, no cap on BM number (median 2, IQR 1-5)
HR 0.83
95% CI 0.53-1.28, p=0.39; Gray's test p=0.66, not met
TL;DRNeurocognitive failure no different for SRS vs HA-WBRT + memantine in SCLC brain mets (adjusted HR 0.83, p=0.39); OS 17.4 vs 8.6 mo favoring SRS.
The RT read is what SRS did not cost: adjusted intracranial failure, neurological death, and salvage therapy were no different from HA-WBRT + memantine, so omitting whole-brain coverage carried no measured CNS penalty. The neurocognitive advantage that justifies SRS in other histologies did not appear here (adjusted HR 0.83, 95% CI 0.53-1.28).
In SCLC with a limited number of brain mets (trial median 2, IQR 1-5), this supports SRS as an alternative to HA-WBRT + memantine; pts with many BM were eligible (range up to 26) but sparsely represented, so the read is thinner there.
The RT read is what SRS did not cost: adjusted intracranial failure, neurological death, and salvage therapy were no different from HA-WBRT + memantine, so omitting whole-brain coverage carried no measured CNS penalty. The neurocognitive advantage seen in other histologies did not appear (adjusted HR 0.83, 95% CI 0.53-1.28).
10 details 2 trials watching
Phase III, randomized 1:1, 151 pts assigned February 2021 to June 2026. Stratified by DS-GPA and number of BM. Median follow-up for surviving pts 9.5 months.
SCLC brain metastases, no upper limit on BM number (initially 10, later amended). Median BM 2 (range 1-26, IQR 1-5). Size and volume thresholds (5 mm from the hippocampi, 4 cm largest diameter, 30 cm total volume) print without inequality symbols in the portal text.
SRS vs HA-WBRT + memantine. Dose, fractionation, and SRS platform are not reported in source.
Primary: time to neurocognitive failure (HVLT-R, COWA, Trail Making), cumulative incidence with death without failure as a competing risk. Secondary: OS, intracranial progression, neurological death, salvage therapies, AEs. Powered at 85% for a 17% absolute reduction with 91 events.
Primary not met; OS favored SRS. Arm-level figures are in the table. CNS control endpoints and G3-5 treatment-related AEs showed no difference in adjusted analyses, with no effect sizes given.
| Endpoint | SRS | HA-WBRT + memantine | Effect (SRS arm) |
|---|---|---|---|
| Neurocognitive failure (1°) | per-arm rate not reported | per-arm rate not reported | adjusted HR 0.83 (95% CI 0.53-1.28), p=0.39 |
| Median OS | 17.4 mo | 8.6 mo | adjusted HR 0.60 (95% CI 0.39-0.91), p=0.016 |
Historical SRS vs WBRT trials excluded SCLC, which left WBRT as the default here. NRG CC001 set HA-WBRT + memantine as the neurocognitive-sparing WBRT standard, so this comparator is the strongest WBRT arm available. FIRE-SCLC supplied the retrospective case for first-line SRS that this trial tests prospectively.
The observed event count is not reported against the 91 events the design required. No per-arm neurocognitive failure rates, no data on systemic therapy or extracranial disease by arm, and no account of what drove the OS gap.
The trial failed its own hypothesis and still weakens the case for WBRT: without a neurocognitive or CNS-control advantage for whole-brain treatment, the argument for it rests on habit. The OS difference is hypothesis-generating, not a tested claim.
Randomised phase III against the WBRT standard: neurocognitive primary not met, but OS and CNS control did not favor WBRT. OS is secondary, f/u 9.5 mo.
- What drives the OS gap when intracranial failure and neurological death were no different?
- Does SRS hold up in SCLC with high BM number? active Stereotactic Radiation in Patients With Small Cell Lung Cancer and 1-10 Brain Metastases Phase NAn=100 · primary completion 2024-04 · SRS instead of WBRT in SCLC with 1-10 BMactive Hippocampal Sparing Whole Brain Radiation Versus Stereotactic Radiation in Patients With 5-20 Brain Metastases: A Phase III, Randomized Trial Phase NAn=196 · primary completion 2024-11 · randomised HA-WBRT vs SRS, 5-20 BM, any solid tumor
- SRS dose, fractionation, and surveillance schedule used
📚 Sources · 📄 1 paper
Abstract
The longer read
The trial was built on a premise borrowed from other histologies: that SRS spares cognition relative to whole-brain treatment. That premise was established against conventional WBRT. NRG-CC009 tested it against the version of WBRT that NRG CC001 had already shown to be gentler on cognition, hippocampal avoidance with memantine, and the neurocognitive gap did not materialize (adjusted HR 0.83, 95% CI 0.53-1.28). One reading is that HA-WBRT + memantine has closed much of the distance. The other is that the trial could not see a real difference. The design assumed a 17% absolute reduction and 91 events, the abstract does not state how many events accrued among 151 pts, and the interval is wide enough to hold a clinically meaningful benefit as well as none. A null primary here is not evidence of equivalence.
The competing-risk structure matters for how to read that null. Neurocognitive failure was analysed with death without failure as a competing event, and the design assumed 29.8% of pts in both arms would die without failing. If survival differed between arms, as the OS result suggests, the arms did not carry the same exposure to cognitive testing. Longer survival gives more time to register a decline, which would work against SRS on this endpoint. The abstract gives no per-arm cumulative incidence, so this cannot be checked from the source.
The OS result (median 17.4 vs 8.6 mo, adjusted HR 0.60, p=0.016) is the number that will travel, and it deserves the most caution. It is a secondary endpoint in a 151-pt trial with 9.5 months of median follow-up in survivors. There is no obvious mechanism by which the choice of brain RT doubles nothing and yet extends survival when intracranial failure and neurological death were no different between arms. Candidate explanations sit outside the abstract: differences in time off systemic therapy, tolerance of subsequent treatment, or imbalance in extracranial disease despite balanced baseline characteristics. None is reported. Until the full paper shows cause of death and systemic therapy delivery by arm, the defensible claim is that SRS was not worse for survival.
What the trial does establish for an RT reader is narrower and more useful. In a population where most pts had few lesions (median 2, IQR 1-5), leaving the uninvolved brain untreated did not produce a measured excess of intracranial failure, salvage therapy, or neurological death in adjusted analyses. That is the prospective, randomised confirmation that FIRE-SCLC could only suggest retrospectively, and it addresses the specific fear behind WBRT in SCLC, that occult disease makes focal treatment futile.
The open edge is burden. The cap on BM number was removed, and the range reached 26, but the interquartile range shows few such pts were enrolled. Technique also gates transfer: the source gives no SRS dose or fractionation, and no detail on surveillance imaging, which is what makes an SRS-first strategy safe in a histology with a high rate of new lesions.