onc brain

About · curated by Nick Boehling, MD · @nb2276
Early signal

RAPCHEM (BOOG 2010-03) NCT01872975

ForcT1-2 (<5cm) cN1 breast cancer, post-neoadjuvant chemo and surgery

10yr locoregional recurrence local control

2.9% (24/838)

Low 2.4%, intermediate 3.2%, high 2.8%

TL;DR10yr locoregional recurrence 2.9% (24/838) with response-adapted RT after neoadjuvant chemo; 2.4% in the RT-omission-eligible low-risk group.

Reported via The ASCO Post →

Why it mattersRadiation oncology

The allocation rule, not the recurrence rate, is the transferable part: ypN0 after mastectomy received no RT at all and still ran 2.4% at 10yr, and ypN1 pts had regional nodes omitted entirely. Dose, fractionation and target-volume detail are not reported in source, which limits direct transfer.

Monday clinic

In cT1-2 cN1 pts who convert to ypN0 after neoadjuvant chemotherapy, this supports the safety of a de-escalated RT volume over 10 years; it does not extend to cN2-3 disease, and the randomised comparison remains open.

8 details 3 trials watching

Prospective multicentre cohort, N=848 across 17 Dutch centres, accrued 2011-2015, presented at EBCC15 with 10-year follow-up; 838 completed follow-up. Not randomised: every patient received the RT volume their risk group assigned.

Breast tumour under 5 cm with 1 to 3 involved lymph nodes at presentation, treated with neoadjuvant chemotherapy then surgery (BCS or mastectomy). Most underwent axillary lymph node dissection.

Volume was set by post-chemotherapy nodal status. ypN0: breast RT after BCS, none after mastectomy. ypN1: breast or chest wall only, regional nodes omitted. ypN2+: breast or chest wall plus regional nodal irradiation. Dose and fractionation are not reported in source.

24 of 838 (2.9%) had a locoregional recurrence without distant spread at 10 years. Per-group counts appear in the detail table; the rates do not separate across strata.

The randomised test of this exact question is NSABP B-51/RTOG 1304 (NCT01872975), which the investigators expect in about 3 years; until then no trial has randomised ypN0 pts to nodal RT versus omission. Prior nodal-RT evidence (MA.20, EORTC 22922) was built in upfront-surgery populations, so it cannot arbitrate a post-chemotherapy response-adapted rule.

cT1-2 cN1 breast cancer treated with neoadjuvant chemotherapy and surgery, staged with axillary dissection
Does not represent cN2-3 disease, tumours over 5 cm, or pts staged by sentinel node biopsy alone.

The 2.9% rate is uninterpretable without a comparator: a low event count in a de-escalated cohort is equally consistent with the omitted RT having been unnecessary and with the cohort being low-risk to begin with. ALND staging also means the ypN0 label carries more information than a modern sentinel-node ypN0 does.

The finding that recurrence is flat at 2.4% / 3.2% / 2.8% across escalating risk is the intended signal: the added RT in the higher strata may be doing the work that keeps them level with the low-risk group. It does not settle whether the low-risk group needed any RT, only that the allocation rule did not produce a visible failure.

Single-arm prospective cohort with no randomised comparator; allocation used ALND-era nodal staging. Confirmatory randomised answer (NSABP B-51) still pending.

📚 Sources · 📄 1 paper
📄 PAPER · The ASCO Post
Breast Cancer Recurrence Remains Low—Even After 10 Years—With Radiotherapy Tailored to Patient’s Individual Risk
Abstract
“The results of our study show that tailoring the extent of radiotherapy according to how well the chemotherapy has worked to treat cancer in the lymph nodes leads to very low and reassuring recurrenc...
📝 Breast Cancer Recurrence Remains Low Even After 10 Years With Radiotherapy Tailored to Patient’s Individual Risk - The ASCO Post

The longer read

The number this study is built around, 2.9% isolated locoregional recurrence at 10 years, is reassuring and, on its own, close to uninterpretable. A single-arm cohort in which every patient received the volume their risk score assigned cannot tell you what would have happened had they received more. That is not a pedantic objection here: the whole clinical question is whether the omitted regional nodal irradiation in the ypN0 and ypN1 groups was carrying any benefit, and a design with no comparator arm is structurally incapable of answering it. What the study does establish is weaker but not trivial, namely that a response-adapted allocation rule, applied prospectively across 17 centres and followed for a decade, did not produce a visible failure anywhere in the system.

The more interesting internal observation is that the three strata are flat: 2.4%, 3.2%, 2.8%. Risk stratification is supposed to identify patients at different baseline risk, and if it worked, an untreated-differential cohort would show a gradient. Flatness is consistent with two very different stories. In one, the escalating RT in the intermediate and high groups is doing real work and has pulled those patients down to the low-risk group's floor, in which case the volumes matter and the rule is correctly calibrated. In the other, all three groups were low-risk after effective systemic therapy and the RT differences are close to irrelevant, in which case even the intermediate group's breast-only treatment may be more than needed. A single-arm design cannot separate these, and the paper's own framing, that recurrence stays low when RT is tailored, is compatible with both.

The generalisability caveat the investigators raise themselves deserves more weight than a footnote. Most patients underwent axillary lymph node dissection, and a ypN0 established by full dissection is a substantially more confident negative than a ypN0 established by sentinel node biopsy, which is what current practice would produce. Anyone importing this allocation rule into a sentinel-node-staged clinic is applying a de-escalation that was validated under better staging information than they have. That is a real transfer risk, and it cuts specifically against the most attractive part of the result, the mastectomy patients who received no radiotherapy at all.

The source also gives no dose, no fractionation, no target-volume definitions, no distant recurrence data, no survival and no toxicity. For a de-escalation study the toxicity side is not decorative: the entire argument for omitting regional nodal irradiation is that you avoid its late costs, and none of that benefit is quantified here. Without it the study reports only half of its own trade-off.

NSABP B-51/RTOG 1304 is the trial that will actually settle this, randomising the ypN0 population to regional nodal irradiation or not, and the investigators put it about three years out. Until then, RAPCHEM's honest position is supporting evidence for a practice many centres have already adopted, rather than the justification for adopting it.