onc brain

About · curated by Nick Boehling, MD · @nb2276

ASTRO Annual Meeting 2026

4 studies Subscribe via RSS →

Challenges SOC

SWOG S1827/MAVERICK

ForLS- or ES-SCLC, no brain mets on MRI after 1st-line therapy

Cognitive failure-free survival safety

HR 0.60

90% CI 0.46-0.78, one-sided p=0.0005, favoring MRI alone

TL;DRMRI surveillance alone beat MRI + PCI for cognitive failure-free survival, HR 0.60 (90% CI 0.46-0.78), in LS- and ES-SCLC without brain mets after first-line therapy.

Why it mattersRadiation oncology

The benefit held with HA-PCI in 77% of the PCI arm and in the HA-PCI-only subgroup, so hippocampal sparing does not rescue PCI. Stage did not modify the effect (LS HR 0.59, ES HR 0.65, interaction p=0.79), which moves the LS-SCLC PCI default toward omission with MRI surveillance, pending final OS.

Monday clinic

In LS- or ES-SCLC with a clean brain MRI after first-line therapy, this supports MRI surveillance over PCI, including HA-PCI. It does not address pts who cannot reliably get serial MRI, and it does not settle OS until the 190-event analysis. The longer read

Also covered Sep 14

SWOG S1827/MAVERICK
Arm6-mo CFFS % (90% CI)12-mo CFFS %
MRI alone37.8 (30.0-45.5)17.3
MRI + PCI16.5 (10.7-23.4)5.9
+2 more figures
Schema: PCI 25 Gy / 10 FX + MRI vs MRI alone; total accrual 304. MRI + cognitive testing at 3, 6, 9, 12, 18, 24 months.
Schema: PCI 25 Gy / 10 FX + MRI vs MRI alone; total accrual 304. MRI + cognitive testing at 3, 6, 9, 12, 18, 24 months.
SWOG S1827/MAVERICK
13 details 3 trials watching

Randomized phase III (SWOG S1827), total accrual 304. Stratified by LS vs ES stage, immunotherapy (y/n), and PS 0-1 vs 2.

SCLC, limited or extensive stage, no prior brain mets and no brain mets on MRI after first-line therapy. 68% LS-SCLC; 41% received immunotherapy with upfront treatment.

PCI 25 Gy / 10 fx, with HA-PCI allowed at physician discretion; 77% of PCI pts received HA-PCI. At brain mets, RT was recommended in both arms (WBRT or SRS allowed).

Primary: cognitive failure-free survival (CFFS), with MRI and cognitive testing at 3, 6, 9, 12, 18 and 24 mo. Key secondary: OS.

CFFS favored MRI alone, HR 0.60 (90% CI 0.46-0.78), one-sided p=0.0005. The effect was similar by stage (LS HR 0.59, ES HR 0.65, interaction p=0.79) and in the HA-PCI-only subgroup. Preliminary OS after 128 events showed no significant difference. PCI decreased brain mets with no PFS difference.

SubgroupCFFS HR
LS-SCLCHR 0.59
ES-SCLCHR 0.65
Interactionp=0.79

The Takahashi Japanese phase III had already moved ES-SCLC toward MRI surveillance, while LS-SCLC PCI still rests on pre-MRI-era randomized data and meta-analysis. MAVERICK is the first randomized test of surveillance vs PCI to include LS-SCLC with modern MRI staging.

LS- and ES-SCLC pts with a negative brain MRI after first-line therapy who can undergo serial MRI
Does not represent pts without MRI access, or pts with brain mets at restaging.

The endpoint uses a 90% CI and a one-sided test. The brain mets reduction with PCI is unquantified in source, as is symptomatic morbidity from salvage RT. ADRIATIC changed the LS-SCLC backbone mid-enrollment.

When OS does not differ, a composite of death + cognitive failure mainly measures cognitive toxicity, and on that axis PCI loses even with hippocampal sparing. The open question is whether final OS confirms the preliminary read.

Randomized phase III, prespecified primary met, diverges from PCI as standard in LS-SCLC. Held below practice-changing: OS immature (128/190 events), composite cognitive endpoint, abstract-only.

📚 Sources · 🐦 1 tweet
Caveats dominate

PACE-A

ForLocalised low/intermediate-risk PCa (GG1-2), surgical candidates, no ADT

TL;DR≥1 urinary pad 14/29 (48%) RP vs 3/36 (8%) SBRT; BCF HR 0.48 (95% CI 0.16-1.46), mFU 8 yrs, GG1-2 localised PCa.

Why it mattersRadiation oncology

The continence gap is the RT read: ≥1 pad in 14/29 (48%) after RP vs 3/36 (8%) after 36.25 Gy/5fx, with bowel toxicity 0 (0%) on SBRT. With mFU 8 yrs and 7 deaths none from PCa, this supports SBRT as the lower-urinary-morbidity option when counselling surgical candidates, though efficacy parity rests on 13 events.

Monday clinic

In an MRI-staged GG1-2, PSA ≤20 surgical candidate choosing between RP and SBRT, this supports framing continence as the main differentiator; it does not extend to GG3+ disease or pts needing ADT. The longer read

PACE-A
Arm≥1 urinary pad
Prostatectomy14/29 (48%)
SBRT3/36 (8%)
+3 more figures
PACE-A
PACE-A
ArmModerate/big bowel problem
Prostatectomy1/28 (4%)
SBRT0 (0%)
SBRT: 36.25 Gy, 5 fractions. 1:1 randomised vs surgery.
SBRT: 36.25 Gy, 5 fractions. 1:1 randomised vs surgery.
12 details

1:1 randomised RP vs SBRT, n=123 (60 RP, 63 SBRT), 10 UK centres. Median follow-up 8 yrs. Phase not stated in source.

Localised low or intermediate risk PCa with surgical consideration: T1c-T2c, Gleason ≤3+4, PSA ≤20, MRI staged, no ADT.

SBRT 36.25 Gy in 5 fractions. Target volume and margins not reported in source.

Biochemical or clinical failure: 13 events, HR 0.48 (95% CI 0.16-1.46), log rank p=0.18. Absolute 5-yr difference 3.1% (90% CI -2.6, 5.0), estimated from HR. 7 deaths, none from PCa, HR 0.55 (0.12-2.46).

Patient-reported ≥1 urinary pad 14/29 (48%) RP vs 3/36 (8%) SBRT. Moderate/big bowel problem 1/28 (4%) RP vs 0 (0%) SBRT.

PACE-B tested SBRT against conventional RT; PACE-A supplies the surgical comparison in the same programme. ProtecT remains the reference randomised comparison of RP vs RT in localised disease, with a different RT technique and population.

MRI-staged T1c-T2c, Gleason ≤3+4, PSA ≤20 PCa in surgical candidates treated without ADT
Does not represent GG3+ or high-risk disease, or pts receiving ADT.

PRO denominators (29 and 36) sit well below the 60 and 63 randomised, so responder bias could inflate or shrink the pad gap. Timepoint for the pad comparison not stated in source.

The trial answers the toxicity question more convincingly than the efficacy one: the urinary continence difference is large, while the BCF CI spans a threefold range either way. Oncologic equivalence of SBRT vs RP in GG1-2 is not settled by these data.

CONSORT flow
Randomized 123
↓
Prostatectomy
allocated 60
SBRT
allocated 63

Randomised but small: 13 BCF events cannot establish oncologic equivalence. Primary endpoint not stated in source; PRO denominators show heavy attrition.

  • Oncologic noninferiority of SBRT vs RP in GG1-2 disease
  • Durability of the continence gap with complete PRO capture
📚 Sources · 🐦 1 tweet
Early signal

STOP

ForPSMA PET ≤3 mets, CI-negative BCR or de novo oligometastatic HSPC

TL;DR3-yr biochemical progression 21% vs 60% (p=0.004) adding SBRT to intermittent ADT in PSMA PET-detected, conventional-imaging-negative oligometastatic HSPC/BCR.

Why it mattersRadiation oncology

The RT read is durability of the ADT-free interval: consolidating all PSMA-avid sites during a fixed 8-9 month ADT course left median time to biochemical progression not reached vs 27 mo. ADT re-initiation trended the same way (p=0.06), so SBRT's value here is extending time off ADT, not replacing it.

Monday clinic

In CI-negative, PSMA PET-positive recurrent or de novo HSPC with ≤3 deposits already planned for a finite ADT course, this supports adding SBRT to prolong biochemical control; it does not address MDT without ADT or disease visible on conventional imaging.

Cumulative incidence of biochemical progression, ADT vs ADT + SBRT, p = 0.004
Cumulative incidence of biochemical progression, ADT vs ADT + SBRT, p = 0.004
+2 more figures
Cumulative incidence of re-initiation of ADT, ADT vs ADT + SBRT, p = 0.06
Cumulative incidence of re-initiation of ADT, ADT vs ADT + SBRT, p = 0.06
STOP
9 details 4 trials watching

Randomised 1:1, n=60. Primary endpoint was feasibility, met with 91% of eligible pts enrolled. Phase not stated in source.

Biochemically recurrent prostate cancer (or de novo metastatic) with negative conventional staging (CT, bone scan) and ≤3 deposits on PSMA PET (miN1, miM1). PSA >0.1 ng/mL after RadP or ≥2 after radical RT.

Both arms received 8-9 months ADT, then stopped. Intermittent ADT is the backbone and the ADT-free interval is the tested space.

SBRT to all PSMA-avid sites per institutional protocol by anatomic region, delivered within 4 months of ADT. No uniform dose/fractionation reported in source.

Biochemical progression favored SBRT (p=0.004); ADT re-initiation trended lower but was not significant (p=0.06). QoL (EORTC QLQ-C30 global health) was collected; values not legible in source.

EndpointiADT + SBRTiADT alonep
3-yr biochemical progression21%60%0.004
Median time to biochemical progressionNR27 monot reported in source
Cumulative incidence of ADT re-initiationnot reported in sourcenot reported in source0.06

STOMP and ORIOLE tested MDT without continuous ADT in conventionally or PSMA-staged oligorecurrence. STOP instead layers SBRT onto a finite ADT course in CI-occult, PSMA-only disease, a population neither directly addressed.

PSMA PET-detected, conventional-imaging-negative oligorecurrent or de novo HSPC with ≤3 deposits on a finite ADT course
Does not represent castration-resistant disease, >3 lesions, or MDT without ADT.

Feasibility-powered, so the efficacy effect size is imprecise and vulnerable to imbalance. Heterogeneous SBRT protocols prevent a dose recommendation, and hard endpoints (MFS, OS) are not reported in source.

The signal is that PSMA-guided consolidation makes a short ADT course more durable. What it does not settle is whether that delays metastatic progression or reduces lifetime ADT exposure, the endpoint that trended but missed.

Randomised but n=60 with a feasibility primary; efficacy is secondary, SBRT unstandardized, and ADT re-initiation not significant. Needs a powered trial.

📚 Sources · 🐦 1 tweet
Challenges SOC

ROADS NCT04365374

ForNewly diagnosed brain met 2-7 cm indicated for resection

Time to surgical bed recurrence local control

HR 0.06

95% CI 0.01-0.46, p=0.0070; co-primary SB-RFS HR 0.48 (0.30-0.76)

TL;DRCs-131 tile brachytherapy at resection cut surgical bed recurrence vs post-op SRT: time-to-SBR HR 0.06 (0.01-0.46), SB-RFS HR 0.48, in newly diagnosed resected brain mets.

Why it mattersRadiation oncology

The surgical bed effect held in the per-protocol population (time-to-SBR HR 0.06), so it is not just an artifact of the 17.8% of SRT pts who never got cavity RT. Control was modern, mostly fractionated 27-30 Gy SRT. This moves the cavity-RT decision from a delayed post-op SRT course toward intra-op Cs-131 implant.

Monday clinic

In a pt with a newly diagnosed 2-7 cm brain met going to resection, this supports discussing intra-op Cs-131 tiles over planned post-op SRT; it does not settle the OS question, or whether TBRT should be chosen over pre-operative SRT. The longer read

Also covered May 30

12 details 1 trial watching

Randomized, open-label, non-inferiority phase 3, 32 US centers, pre-operative 1:1 randomization. NI null HR 1.26 with hierarchical testing (NI then superiority for both co-primaries, then OS). Median f/u 12.9 mo.

One surgical brain metastasis 2.0-7.0 cm, newly diagnosed. mITT required surgery, pathologic confirmation of metastasis and any follow-up: 204 of 230 randomized. Lung primary was most common index histology.

TBRT: collagen tile with four Cs-131 seeds placed in the cavity at resection, after intra-op path. SRT: cavity 17-20 Gy/1, 27 Gy/3 or 30 Gy/5, planned 21±7 days post-op; 92.2% fractionated. Non-index mets got SRT in both arms.

Co-primary: time-to-SBR (death censored) and SB-RFS (SBR or death). Secondary: OS, QoL, KPS, neurocognition, LMD, RN. SBR centrally reviewed by two neuroradiologists.

Any AE 79.0% vs 80.7%, ≥G3 TRAEs 20.0% vs 21.7%. 12-mo RN 5.3% vs 5.7%. No QoL or neurocognitive cost detected.

Post-op SRT became SOC from Mahajan et al. (12-mo freedom-from-SBR 72% vs 43% with observation) and the SRT vs WBRT trial. The R+SRT arm's 15.4% 12-mo SBR sits within the 84-87% freedom-from-SBR of cited fractionated series, so the control was not a weak comparator.

pts with a newly diagnosed, single resectable 2-7 cm brain metastasis, with or without up to 5 other SRT-treated mets
Does not represent unresected mets, recurrent or previously irradiated cavities, or a head-to-head vs pre-operative SRT.

Co-primary endpoints were redefined mid-trial (May 2025) after a death-censoring error in SB-RFS. Seed MRI artifacts may have unblinded central review. OS benefit (HR 0.59) disappeared in PP (HR 0.73, 0.44-1.20) and systemic therapy was not captured in detail.

The local-control gain is large and consistent across mITT, ITT and PP, and plausibly reflects both higher cavity BED and removing the post-op gap in which 18 SRT pts dropped out. The OS signal is hypothesis-generating: confined to single-met pts, sensitive to population, with no clear mechanism.

CONSORT flow
Assessed / enrolled 240
↓ 10 excluded
Randomized 230
↓
R+TBRT
allocated 115
analyzed 103
Median time-to-SBR not reached
R+SRT
allocated 115
analyzed 101
Median time-to-SBR 17.4 mo

Randomized phase 3, co-primaries met on non-inferiority and superiority vs contemporary SRT. Held back from practice-changing by 12.9-mo f/u, mid-trial endpoint amendment, open-label design.

📚 Sources · 📄 1 paper
📄 PAPER Weinberg, Jeffrey S.; Imber, Brandon S.; DiNapoli, Vincent et al. · Journal of Clinical Oncology (2026-09)
Surgery and tile-based radiation therapy versus surgery and stereotactic radiation for newly diagnosed brain metastases (ROADS): a randomized, open-label, phase 3 trial
Abstract
PURPOSE Post-operative stereotactic radiation (SRT) is the standard-of-care for resected brain metastases. Implantation of cesium-131 collagen tiles (tile-based radiation therapy, TBRT) initiates focal radiation immediately after resection, potentially offering therapeutic and logistical advantages. ROADS: a randomized, open-label, non-inferiority, phase 3 trial ( NCT04365374 ) compared the safety and efficacy of resection with TBRT (R+TBRT) to resection with SRT (R+SRT) for patients with a newly diagnosed brain metastasis indicated for surgical resection. PATIENTS AND METHODS Across 32 United States centers, patients were pre-operatively randomized 1:1 to resection R+TBRT or R+SRT. Any non-resected brain metastases received SRT post-operatively. Co-primary outcomes were time-to-surgical bed recurrence (SBR) and surgical bed recurrence-free survival (SB-RFS). Outcomes were analyzed using Cox proportional hazards models with stratification factors as covariates. Multiplicity was controlled by hierarchical testing. Analyses used the pre-specified modified intent-to-treat (mITT) population (patients who underwent surgery, had pathologic confirmation of brain metastasis, and had follow-up information). RESULTS From April 2021 through August 2025, 230 patients were randomized (115 per arm); 204 of whom (103 R+TBRT, 101 R+SRT) comprised the mITT population. Median follow-up was 12.9 months. Median time-to-SBR was not reached (R+TBRT) versus 17.4 months (R+SRT) (hazard ratio [HR]: 0.06; 95% confidence interval [CI]: 0.01–0.46; p=0.0070). SB-RFS was improved with R+TBRT, with a median of not reached versus 10.9 months (R+SRT) (HR: 0.48; 95% CI: 0.30–0.76; p=0.0021). Overall adverse events did not appear to differ: 83 patients (79.0%, 95% CI: 70.0–86.4) (R+TBRT) versus 67 patients (80.7%, 95% CI:70.6–88.6) (R+SRT). CONCLUSION For patients with newly diagnosed brain metastases requiring resection, R+TBRT significantly improved SBR and SB-RFS, demonstrating both non-inferiority and superiority.