MD Anderson Made the Case for Radiation Cleanup After Immunotherapy. Then Its Own Trial Undercut It.

Mehmet Altan stood up at the world’s largest lung cancer meeting this month and did something the field does not do often enough: he presented a result that undercut a strategy his own institution had spent a decade helping to sell. His trial, run at MD Anderson Cancer Center in Houston, had asked a question radiation oncologists thought was already settled. When a patient with metastatic lung cancer responds to immunotherapy, should you send in radiation or surgery to clean up the tumors that are left?

For most of the last decade the assumed answer was yes. The LONESTAR trial is the first randomized, immunotherapy-era test of that assumption, and it did not just fail to confirm the old story. It made a decade of confidence look premature. The patients who skipped the extra treatment did at least as well as the ones who got it, and in the group the strategy was built for they did better.

The idea that got ahead of the evidence

The strategy has a clinical name, local consolidative therapy, and a seductive logic behind it. Cancer that has spread to only a handful of sites, so-called oligometastatic disease, was supposed to be different in kind from cancer that has seeded everywhere. Blast or cut out the few visible deposits, the thinking went, and you buy the patient real time.

It was not a fantasy to start with. A small randomized phase II trial led by Daniel Gomez, also out of MD Anderson, reported in 2019 that oligometastatic lung cancer patients who received local therapy lived a median of 41.2 months against 17.0 months for those who did not. That gap is enormous, and it launched a thousand practice patterns. Around it accumulated a wall of supporting literature, most of it retrospective: a 2025 analysis in Scientific Reports reporting that consolidation “extends survival” after first-line immunotherapy regardless of PD-L1 status, a propensity-matched multicenter series in Lung Cancer, this one in EGFR-mutant disease on targeted therapy rather than immunotherapy, a long-term outcomes abstract at ASCO this year. Each one pointed the same direction. None of them randomized anybody.

That is the whole problem, because retrospective data has a built-in flattery problem. The patients a radiation oncologist offers aggressive local treatment to are, as a rule, the ones already doing well: good performance status, limited disease, organs that can take the beam. They were going to outlive the sicker patients who were never candidates, radiation or no radiation. Give the beam the credit for their survival and you have mistaken the selection for the treatment. The only way to break that confound is to flip a coin, and Gomez’s 2019 trial flipped it for just 49 patients, in the chemotherapy era, before modern immunotherapy rewrote what systemic control even means.

What the coin flip actually showed

LONESTAR flipped it again, larger and later. After twelve weeks of induction with nivolumab plus ipilimumab, Bristol Myers Squibb’s Opdivo and Yervoy, the 166 patients who had not progressed were randomized: 83 continued the drugs alone, 83 got local consolidative therapy first and then continued. Seventy-one of the treated patients received radiation to at least one site; sixteen went to surgery. Seventy-seven had oligometastatic disease, the exact population the whole premise was built for.

Overall survival ran 52.8 months in the drugs-only arm and 43.2 months in the group that got the extra local treatment, a hazard ratio of 1.14 that did not come near statistical significance. Progression-free survival tilted the other way, 31.3 months with local therapy against 24.3 without, and that difference did not clear the bar either. The radiation did its clean-up job on the scans and did not add a day of life anyone could measure.

Then the subgroup that was supposed to benefit most. In the oligometastatic patients, median overall survival was 75.8 months for those who took the drugs alone and 42 months for those who added local therapy. In that same subgroup the drugs-alone patients also stayed progression-free longer, 44.0 months against 35.7. This is a subgroup of a negative trial, small and underpowered, and no honest reading calls it proof of harm. But for a treatment that was supposed to shine brightest right there, both needles pointed the wrong way.

Why the theory broke

The most plausible explanation is not exotic once you notice what modern immunotherapy does and what radiation does to it. Dual checkpoint blockade now delivers durable, body-wide disease control on its own; a 52.8-month median survival in metastatic lung cancer would have read as science fiction fifteen years ago. When systemic control is that good, the results suggest, there is little residual disease left for local therapy to consolidate, and the downside stops being theoretical.

LONESTAR measured that downside. The blunt safety tally did not move much: local therapy did not raise the overall rate of grade 3 or worse adverse events. Two more specific signals did, and both bear on how the drugs work. Pneumonitis, inflammation of the lungs, ran 9.5% in the local-therapy arm against 4.9% with the drugs alone. And when patients restarted systemic treatment after radiation, investigators recorded markedly lower absolute lymphocyte counts. Those lymphocytes are the T-cells that checkpoint inhibitors exist to unleash, and radiation does not spare them. The lower counts in the treated arm are the fingerprint of a beam aimed at the tumor also thinning the immune army the expensive drugs had just mobilized. In the chemotherapy era, when systemic control was weak, mopping up locally made sense. In the immunotherapy era, the trade reads as toxicity and lost lymphocytes bought against a benefit that is no longer there.

None of this makes local radiation useless in lung cancer. It remains standard for palliating symptoms and for disease that breaks through in a single spot. What LONESTAR punctures is the broader claim, the one that migrated out of small trials and retrospective series into routine practice: that reflexively consolidating everything after a good immunotherapy response extends life. A companion viewpoint in JAMA Oncology this year was already asking where the randomized evidence for the habit actually sat. Now there is an answer, and it is not the one the field spent a decade assuming.

Altan’s trial was run at a single center, and one negative study does not close a question that a dozen enthusiastic ones opened. But sit with the shape of what happened. The institution that handed oncology its most-quoted argument for blasting leftover tumors is the same one that built the trial to test it, and reported the null without flinching. That is how the record is supposed to correct itself. It just took a randomized trial, and the better part of a decade, to catch up to a coin flip.

Sources

1. News-Medical – MD Anderson trial challenges local consolidative therapy paradigm in NSCLC (LONESTAR, WCLC 2026)
2. MedPage Today – Local Consolidative Therapy After Dual Immunotherapy Fizzles in Metastatic NSCLC
3. Medical Xpress – Local consolidative therapy does not improve survival after dual immunotherapy in metastatic NSCLC
4. Journal of Clinical Oncology – Gomez et al., local therapy vs maintenance/observation in oligometastatic NSCLC, long-term results (2019)
5. Scientific Reports – LCT extends survival in metastatic NSCLC with oligoresidual disease after first-line immunotherapy (2025)
6. Journal of Clinical Oncology – Long-term survival with consolidative local ablative therapy in oligometastatic NSCLC (ASCO 2026 abstract)
7. Lung Cancer – Clinical impact of LCT in EGFR-mutant metastatic NSCLC: propensity-matched multicenter analysis (2026)
8. JAMA Oncology – Local Consolidative Therapy in Lung Cancer (2026)