Why this matters
SABR provides excellent local control for early-stage node-negative NSCLC, but regional relapse, distant metastasis, and second primary lung cancers remain important causes of treatment failure.
The I-SABR trial tested whether a short course of immunotherapy around SABR could reduce those failures and ultimately improve long-term survival.
Earlier results established an event-free survival advantage. The ASTRO 2026 update now provides the more consequential result: after more than 6 years of follow-up, the separation extends to overall survival.
Study design
I-SABR was an open-label randomized phase II trial enrolling patients with biopsy-confirmed, node-negative NSCLC suitable for SABR.
Eligible patients included:
- newly diagnosed early-stage NSCLC
- isolated lung-parenchymal recurrent or persistent NSCLC suitable for SABR
Patients were randomized to:
- SABR alone
- SABR plus nivolumab
Nivolumab was given at 480 mg every 4 weeks for 4 doses, beginning before or within 36 hours of the first SABR fraction.
The primary endpoint was 4-year event-free survival.
Events included:
- local recurrence
- regional recurrence
- distant metastasis
- second primary lung cancer
- death
Overall survival was a secondary endpoint.
A total of approximately 140 patients were randomized.
Median follow-up in the current presentation was 74.5 months.
Key results
The long-term overall survival curves separated in favor of I-SABR.
At 6 years:
- OS: 69% with I-SABR vs 55% with SABR
- log-rank p=0.0466
- Cox HR approximately 0.55
Updated event-free survival also remained significantly better:
- 6-year EFS: 55% with I-SABR vs 32% with SABR
- HR approximately 0.52
- p approximately 0.005
The magnitude of effect was therefore substantial for both long-term disease control and survival.
Failure patterns also favored I-SABR.
Compared with SABR alone, I-SABR reduced the combined occurrence of local recurrence, distant metastasis, and second primary lung cancer by approximately 70%.
Selected first failure rates included:
- same-lobe failure: 4.5% vs 16%
- distant metastasis: 4.5% vs 17.3%
- second primary lung cancer: 4.5% vs 16%
Regional nodal failure was similar between the groups.
Subgroup findings
The survival benefit generally favored I-SABR across the examined clinical subgroups.
The presentation suggested benefit even among patients who might traditionally be considered lower risk, including those with smaller tumors and negative PD-L1 expression.
However, these subgroup analyses were not individually powered and should not be used to define treatment selection on their own.
Translational findings
The presentation also included immune-correlative data supporting a potential biological interaction between SABR and nivolumab.
I-SABR was associated with sustained CD4-positive helper T-cell related cytokine activity and longitudinal expansion of activated cytotoxic CD8-positive T-cell clonotypes.
Patients treated with I-SABR who remained recurrence-free demonstrated more persistent expansion of activated CD8-positive T-cell populations, whereas patients who recurred and those receiving SABR alone showed less clonal expansion.
These findings are mechanistically interesting, but they remain exploratory and should not be treated as validated biomarkers for selecting patients for immunotherapy.
Interpretation
The most important result is the durability of the treatment effect.
I-SABR was initially attractive because it improved event-free survival. With more than 6 years of follow-up, the benefit has not disappeared. Instead, the curves remain separated and now show an overall survival difference of approximately 14 percentage points.
That makes the result substantially more clinically interesting.
The treatment strategy is also notably limited in duration. Patients received only four doses of nivolumab rather than a year or more of adjuvant immunotherapy.
The failure pattern provides another important clue. The benefit was not confined to one mode of recurrence. I-SABR reduced local failure, distant metastasis, and second primary lung cancers, while regional nodal failure remained similar.
This pattern is compatible with a systemic immune effect rather than simply improved local treatment.
However, the strength of the conclusion needs to match the design.
This remains a randomized phase II trial with approximately 140 patients, and overall survival was a secondary endpoint. The OS significance is also statistically borderline, with a log-rank p-value just below 0.05 and a Cox confidence interval reaching approximately 1.00.
The magnitude and duration of the effect are impressive, but this is not the same level of evidence as a large confirmatory phase III trial.
I-SABR therefore provides a strong proof-of-concept signal that short-course immunotherapy can improve long-term outcomes after SABR in node-negative NSCLC, but it does not yet establish four doses of nivolumab as a universal new standard for all patients receiving SABR.
Limitations
This was a relatively small randomized phase II trial.
Overall survival was a secondary endpoint rather than the original primary endpoint.
The OS result is statistically borderline despite a clinically meaningful absolute difference.
The population was heterogeneous, including both treatment-naive early-stage disease and isolated parenchymal recurrence.
The trial was open-label.
Several subgroup analyses are underpowered and should be considered exploratory.
The immune-correlative analyses are hypothesis-generating and do not yet provide a validated method for selecting patients who need immunotherapy after SABR.