Background
Pooled analysis of two prematurely closed phase 3 RCTs comparing SABR vs lobectomy in operable stage I NSCLC: STARS (MD Anderson; N=40 enrolled before closure) and ROSEL (Netherlands Cancer Institute; N=18 enrolled). Both trials closed early due to poor accrual. Combined N=58 patients (31 SABR, 27 surgery). Operable defined as clinical stage IA–IIA, surgical candidate.
Interventions and follow up
Arm A: SABR: 54 Gy/3 fractions (peripheral) or 50 Gy/4 fractions (central/ultracentral)
Arm B: Lobectomy with mediastinal lymph node dissectio
Primary endpoint: 3-year OS
mFollow up: Median 40.2 months (SABR), 35.4 months (surgery)
Arm B: Lobectomy with mediastinal lymph node dissectio
Primary endpoint: 3-year OS
mFollow up: Median 40.2 months (SABR), 35.4 months (surgery)
Results
3-yr OS: 95% (SABR) vs 79% (surgery), HR 0.14 (95% CI 0.017–1.19), P=.037
3-yr RFS: 86% vs 80%, HR 0.69 (P=.54)
Grade ≥3 toxicity: 10% SABR vs 44% surgery, P=.0059
Grade 5 events: 0 SABR vs 4 surgery
3-yr RFS: 86% vs 80%, HR 0.69 (P=.54)
Grade ≥3 toxicity: 10% SABR vs 44% surgery, P=.0059
Grade 5 events: 0 SABR vs 4 surgery
Adverse events
Main adverse events: Grade ≥3 toxicity substantially lower with SABR (10% vs 44%, p=0.0059). Surgery arm: grade 5 events in 4 patients (1 intraoperative death, 3 from complications — respiratory failure, empyema). SABR: no grade 5 toxicity. Chest wall pain and radiation pneumonitis (grade ≤2) most common with SABR.
Conclusions
In this pooled analysis, SABR was associated with superior 3-year OS (95% vs 79%, HR 0.14) and dramatically lower grade ≥3 toxicity compared to lobectomy in operable stage I NSCLC. Results suggest SABR may be comparable or superior to surgery, though extremely small N severely limits definitive conclusions.
Key Limitations
Key Limitations: Extremely small N=58 — only 10 death events total; the trial was massively underpowered for OS as a primary endpoint. Both trials closed early due to poor accrual, introducing selection bias (patients with specific preferences enrolled). Post-hoc pooled analysis of two separately designed RCTs with different eligibility criteria. Surgeons may have selected better-risk operable patients; disease characteristics may differ. Long-term outcomes (5-year) unavailable. No stratification for pathologic N status or grade.
Clinical Context
Provocative results driving the hypothesis that SABR ≈ surgery for operable stage I NSCLC. Multiple larger phase 3 trials are ongoing to definitively answer this question: STABLE-MATES, SABRTooth (UK), VALOR (VA), POSTILV. Until mature RCT data are available, NCCN guidelines recommend surgery as preferred for operable patients. SBRT remains standard for medically inoperable patients. Patient preference discussions regarding SBRT vs surgery are appropriate in multidisciplinary settings.
References