SWOG S1007
ForHR+/ERBB2- breast, 1-3 positive nodes, Oncotype RS ≤25
TL;DR5yr LRR 0.85% with RNI vs 0.55% without after BCS+RT in RS≤25 N1 disease; IDFS unchanged by RNI.
The number that moves the RNI decision is 0.55% 5yr LRR after BCS+RT without RNI, and equally low LRR in the endocrine-alone arm. Chemotherapy omission on a low RS does not by itself justify adding supraclavicular coverage. Target-volume detail beyond supraclavicular is not reported in source.
In HR+/ERBB2- N1 disease with RS ≤25 treated with BCS and whole-breast RT, this argues low RS alone does not compel nodal irradiation; it does not speak to RS >25, higher nodal burden, or ERBB2+ disease.
The actionable number is 0.55% 5yr LRR after BCS+RT without RNI in RS ≤25 N1 disease, against 0.85% with RNI. That floor leaves little absolute room for regional coverage, and IDFS did not move (premenopausal HR 1.03, postmenopausal HR 0.85). This informs the elective supraclavicular decision.
LRR stayed similarly low in the endocrine-therapy-alone group, so omitting chemotherapy on a low Recurrence Score did not raise locoregional risk. The conclusion is explicit: chemo omission is not an independent indication for RNI, which removes a reason to hedge a de-escalation decision.
9 details
Secondary analysis of the randomized SWOG S1007 trial (endocrine therapy alone vs chemotherapy then endocrine therapy). Radiotherapy data were prospectively collected across diverse practice settings, but RNI receipt itself was not randomized. Data analyzed June 2022 to April 2023.
Hormone receptor-positive, ERBB2-negative breast cancer with 1 to 3 involved nodes and Oncotype DX 21-gene Recurrence Score ≤25. 4871 female patients had radiotherapy forms; median age 57 (range 18-87).
RNI defined as targeting at least the supraclavicular region. 3947 (81.0%) reported radiotherapy receipt; of 3852 with complete target information, 2274 (59.0%) received RNI. Dose, fractionation, and internal mammary coverage are not reported in source.
Cumulative incidence of locoregional recurrence by locoregional treatment received, plus association between invasive disease-free survival and locoregional therapy, adjusted for menopausal status, treatment group, recurrence score, tumor size, nodes involved, and axillary surgery.
Median follow-up 6.1 years. See the LRR and IDFS tables above; IDFS did not differ by RNI receipt in either menopausal stratum.
| Locoregional treatment | 5yr LRR |
|---|---|
| BCS + RT with RNI | 0.85% |
| BCS + RT without RNI | 0.55% |
| Mastectomy + PMRT | 0.11% |
| Mastectomy, no RT | 1.7% |
| Group | HR (95% CI) | P |
|---|---|---|
| Premenopausal | 1.03 (0.74-1.43) | .87 |
| Postmenopausal | 0.85 (0.68-1.07) | .16 |
MA.20 and EORTC 22922 established the regional irradiation question in node-positive disease, both without an overall survival gain, in cohorts assembled before genomic risk stratification. This analysis reads that question in the population those trials could not define, biologically favorable N1, and finds an LRR floor low enough that a relative benefit has little absolute room to operate.
Radiotherapy information was recorded only in the first year after randomization, forcing a 1-year landmark that excludes the earliest events. Target detail stops at supraclavicular coverage, so internal mammary treatment cannot be separated, and the low absolute event count leaves the IDFS confidence intervals wide enough to accommodate a small effect in either direction.
The finding that matters is the floor, not the comparison: with 5yr LRR at or below 1.7% in every locoregional strategy examined, the population has too few events for regional irradiation to demonstrate meaningful absolute benefit. The explicit conclusion is that omission of chemotherapy is not an independent indication for RNI, which addresses a specific compensatory reflex rather than the general RNI question.
Prospectively collected RT data in a large trial cohort, but RNI receipt was not randomized; observational comparison, so an unmeasured-confounding read is unavoidable.
- Does any favorable N1 subgroup (3 nodes, RS near 25) still warrant RNI?
- Internal mammary coverage contribution, unseparable in this dataset
- Longer follow-up for late locoregional events in HR+ disease
📚 Sources · 📄 1 paper
Abstract
The longer read
The useful way to read this analysis is not as a test of regional nodal irradiation but as a measurement of how much room is left for it. In a population defined by hormone receptor positivity, ERBB2 negativity, one to three involved nodes, and a 21-gene Recurrence Score of no more than 25, five-year locoregional recurrence sat at 0.85% after breast-conserving surgery with radiotherapy plus RNI, 0.55% after breast-conserving surgery with radiotherapy without RNI, 0.11% after mastectomy with postmastectomy radiotherapy, and 1.7% after mastectomy without radiotherapy. Those are event rates at which the arithmetic of relative risk reduction stops being clinically interesting. A therapy that halves a 0.55% risk returns something the reader cannot detect in a clinic population and cannot weigh against late cardiac, pulmonary, or lymphedema cost.
The comparison to MA.20 and EORTC 22922 is where the interpretive work sits. Both established regional irradiation as a defensible option in node-positive disease, both did so without an overall survival gain, and both assembled their cohorts before genomic assays existed to separate biologically indolent node-positive disease from the rest. Their control-arm recurrence rates were not this low. What this analysis contributes is the observation that the population those trials studied has since been fractionated, and the fraction defined by a low Recurrence Score carries a locoregional risk profile the original trials never isolated. Applying an effect size measured in a mixed-biology cohort to the favorable slice of it is precisely the extrapolation this data set argues against.
The methodological constraint is real and cuts in a specific direction. RNI receipt was not randomized, so the comparison between irradiated targets is observational and confounded by exactly the features that drive the decision: nodal burden within the one-to-three range, extranodal extension, age, and institutional practice. That confounding should bias toward making RNI look worse, since patients with more concerning features preferentially received it, and the observed 0.85% versus 0.55% is consistent with that direction. It does not, however, establish that RNI provides nothing; the study cannot answer that question and does not claim to. The invasive disease-free survival hazard ratios of 1.03 in premenopausal and 0.85 in postmenopausal patients carry confidence intervals wide enough to contain modest benefit or modest harm, and the one-year landmark imposed by the timing of radiotherapy data collection removes the earliest events from view.
The conclusion worth carrying is narrower and better supported than a general statement about RNI. The authors address a specific reflex: when chemotherapy is omitted on the strength of a low Recurrence Score, some clinicians compensate by intensifying local therapy. The data show locoregional recurrence was similarly low in the endocrine-therapy-alone group, which means the compensatory logic has no empirical support. That is a defensible claim from an observational data set, because it does not require the RNI comparison to be unconfounded. It requires only that the endocrine-alone group's absolute event rate be low, and it is. What remains genuinely open is whether any identifiable subgroup within favorable N1 disease, perhaps three involved nodes with a Recurrence Score near 25, carries enough risk to change the calculation. This analysis does not resolve that, and the target-volume detail it reports, which stops at supraclavicular coverage, would not let it.