Tile-Based Radiation Improves Local Control in Resected Brain Metastases

Resection plus cesium-131 tile-based radiation therapy (R+TBRT; GammaTile) significantly improved time to surgical bed recurrence (SBR) and SB recurrence-free survival (SB-RFS) vs resection plus stereotactic radiation therapy (R+SRT) in patients with newly diagnosed brain metastases, according to results from the phase 3 ROADS trial (NCT04365374) published online in the Journal of Clinical Oncology.^1,2 The full analysis was previously presented as a late-breaking abstract at the 2026 American Society of Clinical Oncology Annual Meeting.¹

Median time to SBR was not reached with R+TBRT vs 17.4 months with R+SRT (HR, 0.06; 95% CI, 0.01-0.46; P = .0070), and median SB-RFS was not reached vs 10.9 months (HR, 0.48; 95% CI, 0.30-0.76; P = .0021). Noninferiority against a prespecified null HR of 1.26 was met for both co-primary end points before superiority was tested.

Study Design

ROADS enrolled patients at 32 US centers from April 2021 to August 2025. Eligible patients had 1 surgical brain metastasis measuring 2.0 to 7.0 cm and up to 5 additional metastases of 4.0 cm or smaller that did not require resection; those lesions received SRT after surgery in both arms. Of 230 patients randomized 1:1 before surgery (115 per arm), 204 comprised the modified intent-to-treat population (103 R+TBRT; 101 R+SRT). Median follow-up was 12.9 months.

Cavity SRT was delivered in 1, 3, or 5 fractions starting 21 ± 7 days after surgery. The co-primary end points were time to SBR and SB-RFS, with site-reported recurrences centrally reviewed by 2 independent, board-certified neuroradiologists.

The protocol originally listed SB-RFS as the sole primary end point. In May 2025, investigators recognized that its definition incorrectly treated death as a censoring event. The protocol was amended June 18, 2025, before any outcome data had been reviewed, and the data safety monitoring board recommended reporting both end points.

Efficacy Findings

One patient in the R+TBRT arm had an SBR event vs 12 in the R+SRT arm. At 12 months, the cumulative incidence of SBR was 1.3% (95% CI, 0.1%-6.1%) vs 15.4% (95% CI, 7.1%-26.6%; subdistribution HR, 0.07; 95% CI, 0.01-0.56; P = .012). SB-RFS events occurred in 30 and 45 patients, respectively. Results for both co-primary end points held in the intent-to-treat and per-protocol populations; in the per-protocol population, median SB-RFS was not reached vs 11.8 months (HR, 0.57; 95% CI, 0.35-0.93; P = .025).

Median overall survival (OS) was 42.5 months (95% CI, 20.8–not estimable) with R+TBRT vs 17.6 months (95% CI, 10.9-25.5) with R+SRT (HR, 0.59; 95% CI, 0.37-0.96; P = .032). The difference was seen in patients with a single brain metastasis (median OS not reached vs 22.1 months; HR, 0.49; 95% CI, 0.25-0.94; P = .033) but not in those with more than 1. The OS finding did not persist in the per-protocol population (HR, 0.73; 95% CI, 0.44-1.20; P = .2156).

Safety and Quality of Life

Any adverse event (AE) occurred in 79.0% of patients with R+TBRT (95% CI, 70.0%-86.4%) vs 80.7% with R+SRT (95% CI, 70.6%-88.6%). Grade 3 or higher treatment-related AEs occurred in 20.0% vs 21.7%. The 12-month cumulative incidence of radiation necrosis was 5.3% (95% CI, 1.7%-12.2%) vs 5.7% (95% CI, 1.8%-13.0%; HR, 0.85; 95% CI, 0.31-2.36; P = .76).

Median duration of craniotomy hospitalization was 3.0 days vs 4.0 days (HR, 1.09; 95% CI, 0.82-1.45; P = .56). Quality of life, functional status, and neurocognitive outcomes did not differ between arms, and time to distant brain failure was similar (HR, 0.93; 95% CI, 0.57-1.54; P = .79). An exploratory composite end point of time to SBR or radiation necrosis favored R+TBRT (HR, 0.28; 95% CI, 0.12-0.66; P = .004).

Clinical Context and Limitations

Among patients in the R+SRT arm, 17.8% (18 of 101) did not receive SRT, most often because of death, withdrawal of consent, or hospice entry. The authors described this as consistent with other prospective trials of postoperative SRT. Per-protocol analyses still favored R+TBRT for time to SBR and SB-RFS.

The trial was open label and was not designed to capture granular data on systemic therapy. The authors wrote that the mechanism behind the OS benefit is unclear, that a plausible explanation is avoidance of delays in starting or resuming systemic therapy, and that this remains speculative. They acknowledged possible unappreciated confounders and noted that the loss of the OS finding in the per-protocol population may reflect immortal time bias, and said the finding requires validation. Baseline characteristics were well balanced, though median cavity diameter differed (3.24 cm with R+TBRT vs 2.70 cm with R+SRT; P = .0090). GT Medical Technologies, which makes GammaTile, sponsored the trial.

The authors also noted that 2 phase 3 trials (NCT03741673; NCT05438212) are evaluating preoperative SRT. They wrote that TBRT is implanted after intraoperative pathologic confirmation of metastatic disease, which avoids treating alternative pathologies.

"R+TBRT should be considered a standard-of-care option for patients with newly diagnosed brain metastases,” concluded Beckham et al.

REFERENCES

1. Weinberg JS, Imber BS, DiNapoli V, et al. Surgery and tile-based radiation therapy vs surgery and stereotactic radiation for newly diagnosed brain metastases (ROADS): a randomized, open-label, phase 3 trial. J Clin Oncol. Published online September 28, 2026. doi:10.1200/JCO-26-01894

2. Post-Surgical Stereotactic Radiotherapy (SRT) Versus GammaTile-ROADS (Radiation One and Done Study). ClinicalTrials.gov. Updated October 7, 2025. Accessed September 30, 2026. https://clinicaltrials.gov/study/NCT04365374