SUPREMO
ForPost-mastectomy pT1-2N1, pT3N0, or pT2N0 grade 3/LVI+ breast cancer
81.4% vs 81.9%
HR 1.04, 95% CI 0.82-1.30, P=0.80; primary endpoint not met
TL;DR10yr OS 81.4% vs 81.9% (HR 1.04, 0.82-1.30, p=0.80): PMRT omission safe in intermediate-risk pN0-pN1 post-mastectomy.
The RT read is the local-control trade: 1.1% vs 2.5% chest-wall recurrence, 29 events total, bought with 40-50 Gy to the chest wall in a population where OS was flat at 10 years. Nodal volumes were not routinely treated (SCF 97/808), so this speaks to chest wall alone, not to regional nodal irradiation.
In a pT2N1 or pT3N0 mastectomy patient who has completed modern adjuvant systemic therapy, this supports discussing PMRT omission with an absolute chest-wall recurrence trade under 2 points; it does not address regional nodal irradiation or pN2-N3 disease.
The trade is 1.1% vs 2.5% chest-wall recurrence from 40-50 Gy, with 10yr OS flat (HR 1.04). Nodal volumes were not routinely treated (SCF 97/808, IMC 12/808), so this licenses chest-wall omission specifically, not regional nodal omission, and moves PMRT here into a morbidity-versus-local-control discussion.
The systemic backbone (85% chemo, 79% endocrine, 19% trastuzumab) is what makes the null interpretable: with modern adjuvant therapy the residual chest-wall event rate is 2.5% untreated, leaving no room for RT to alter survival. Referral for PMRT in this band becomes optional rather than expected.
After mastectomy plus an axillary procedure in pT1-2N1, pT3N0, or pT2N0 grade 3/LVI disease, expected PMRT no longer carries a survival argument (HR 1.04), which changes the reconstruction conversation at the time of surgery since a planned reconstruction need not be sequenced around anticipated chest-wall irradiation.
Also covered Jul 9
9 details
International phase 3 randomized trial (BIG 2-04 MRC/EORTC SUPREMO), 125 UK sites plus 27 European and 21 international sites. N=1607 ITT (808 CWI, 799 no CWI), randomized August 2006 to April 2013, database lock June 2024. Median follow-up 9.6 years.
"Intermediate-risk" post-mastectomy disease: pT1N1, pT2N1, pT3N0, or pT2N0 with grade 3 and/or LVI. All had mastectomy, an axillary procedure, and systemic therapy. Baseline systemic exposure: 85% chemotherapy, 79% endocrine, 19% trastuzumab.
Chest wall 40 to 50 Gy in the irradiation arm. Nodal volumes were not part of the randomized question: supraclavicular fossa treated in only 97/808 irradiated patients, internal mammary chain in 12/808. Twelve patients in the no-irradiation arm received SCF treatment.
Primary: overall survival at 10 years. Secondary: chest-wall recurrence, regional recurrence, disease-free survival, distant metastasis-free survival, cause of death, radiation-related adverse events.
The historic case for postmastectomy RT in node-positive disease rests on the EBCTCG overview, where the locoregional-control gain translated into a mortality benefit. SUPREMO tests that inheritance in the 1-3 node and high-risk node-negative band under contemporary systemic therapy and finds the recurrence signal preserved (HR 0.45) but the survival signal absent (HR 1.04).
The chest-wall recurrence benefit rests on 29 events total with a CI upper bound of 0.99, so the point estimate is unstable. Accrual ran 2006-2013, predating routine dual HER2 blockade, extended adjuvant CDK4/6 inhibition, and current genomic risk stratification, all of which lower the baseline recurrence rate this trial was powered against.
A flat OS with a halved chest-wall recurrence is the signature of a locoregional intervention operating below the threshold where local control converts into survival. At 1.1% vs 2.5%, the absolute chest-wall event rate in both arms is low enough that no plausible salvage-to-mortality pathway could move a 10-year OS curve. The result reframes PMRT in this band as a local-control decision to be weighed against RT morbidity, not as a survival decision.
| Endpoint | CWI | No CWI | HR (95% CI) |
|---|---|---|---|
| Overall survival (1°) | 81.4% | 81.9% | 1.04 (0.82-1.30), p=0.80 |
| Disease-free survival | 76.2% | 75.5% | 0.97 (0.79-1.18) |
| Distant MFS | 78.2% | 79.2% | 1.06 (0.86-1.31) |
| Chest-wall recurrence | 9 (1.1%) | 20 (2.5%) | 0.45 (0.20-0.99) |
CONSORT flow
Adequately powered phase 3, prespecified OS primary, 9.6yr median follow-up, modern systemic backbone. Supports omitting PMRT in a population where guidelines still often recommend it.
- Does regional nodal irradiation carry the same null in pN1 disease
- Which biomarker or genomic subgroup still benefits from chest-wall RT
- Late cardiac and second-malignancy burden of the irradiated arm
📚 Sources · 📄 1 paper
Abstract
The longer read
The central claim in postmastectomy radiotherapy has always been inherited rather than directly tested in this population: the EBCTCG overview established that preventing locoregional recurrence in node-positive disease converts into a breast-cancer mortality reduction, and guidelines extended that logic downward into the one-to-three node band and into node-negative disease carrying grade 3 or lymphovascular invasion. SUPREMO is the trial that asks whether the conversion still happens when the systemic backbone is modern. It does not.
The two headline numbers should be read together rather than in sequence. Chest-wall irradiation did what radiotherapy is supposed to do: chest-wall recurrence fell from 2.5% to 1.1%, hazard ratio 0.45. Overall survival did not move at all, 81.4% versus 81.9%, hazard ratio 1.04 with a confidence interval from 0.82 to 1.30 that comfortably excludes any survival gain of the size guidelines have historically assumed. The mechanism is arithmetic rather than biological. With 29 chest-wall recurrences across 1607 patients, the pool of events available to be converted into deaths is too small for even complete conversion to register on a ten-year survival curve. Radiotherapy did not fail here; the population was chosen such that success could not matter to survival.
That reasoning is what should move a reader's confidence, and it should move it toward accepting the null rather than treating it as a power problem. A trial that observed no local-control effect would leave open the possibility of underdosing or poor technique. A trial that observed the expected local-control effect and no survival effect has isolated the question cleanly. The 9.6-year median follow-up is long enough that late breast-cancer deaths, which dominate here at 65.8% of all deaths, have had time to accumulate.
Two boundaries limit how far this travels. The randomized question was the chest wall, not the regional nodes: supraclavicular fossa treatment occurred in 97 of 808 irradiated patients and internal mammary chain in 12, so this is not evidence about regional nodal irradiation, which carries its own trial base and its own toxicity profile. And the trial's own vintage cuts in a direction that reinforces its conclusion rather than undermining it. Accrual closed in 2013, before dual HER2 blockade, extended adjuvant endocrine intensification, and current genomic stratification became routine. Each of those lowers baseline recurrence risk further, which shrinks the absolute local-control benefit radiotherapy has available to deliver in a patient treated today. If SUPREMO overestimates anything, it overestimates the case for treating.
The residual uncertainty sits in the chest-wall recurrence estimate itself, whose upper confidence bound touches 0.99. A reader should not carry the 0.45 point estimate around as a stable quantity. What survives is the shape of the result: a real but small local-control gain, no survival gain, in a population where the competing risk is distant disease. That converts postmastectomy radiotherapy in this band from a survival decision into a morbidity-versus-local-control conversation, which is a different conversation to have with a patient.