Long-term impact of regional nodal irradiation after neoadjuvant systemic therapy
ForcN+ stage II-III breast, NAT then BCS/mastectomy with ALND
TL;DRAdjusted LRR HR 0.497 and DR HR 0.731 with RNI after NAT in cN+ stage II-III breast; unadjusted 10-yr LRR 9.4% vs 8.1%, P=.62.
The unadjusted 10-yr LRR curves barely separate (9.4% vs 8.1%), yet RNI held up in residual-nodal disease on MVA (HR 0.476), which is where most RNI is already given. The axillary pCR arm stays uninformative: 5.5% vs 4.6% with too few events to model, so it does not answer the omission question.
In cN+ breast with residual axillary disease after NAT and ALND, this supports keeping RNI in the plan; it does not establish safe RNI omission after axillary pCR.
The unadjusted 10-yr LRR curves barely separate (9.4% vs 8.1%), yet RNI held up in residual-nodal disease on MVA (HR 0.476), where nodal coverage is already standard. The axillary pCR group (5.5% vs 4.6%) had too few events to model, so it cannot support omission.
12 details 3 trials watching
Single-institution retrospective cohort, N=1289, 1989-2007. Median follow-up 10.2 yrs after definitive surgery. LRR and DR by cumulative incidence with competing risks; Cox multicovariate models.
Stage II-III breast cancer with cytologically confirmed axillary metastases, NAT, then BCS or mastectomy with ALND. Axillary pCR in 368 (28.5%).
RNI defined as RT purposely directed to at least one regional nodal basin, at physician discretion, given to 1080 (83.8%). Median breast/chest wall 50 Gy/25 fx, median boost 10 Gy/5 fx; institutional standard covered undissected axilla, SCV and IMN, with variable field design. Of 209 without RNI, 142 had no RT at all.
LRR (breast or regional nodes) and DR (LRR or DM, first event), plus DSS and OS.
Univariate RNI effect null for LRR and DR. Adjusted, RNI reduced LRR (HR 0.497) and DR (HR 0.731), with the largest DR effect in HER2+ pts on trastuzumab (HR 0.237). No DSS or OS effect.
| Endpoint | No RNI | RNI | P |
|---|---|---|---|
| LRR, all pts | 9.4% | 8.1% | .62 |
| LRR, axillary pCR | 5.5% | 4.6% | .98 |
| LRR, no axillary pCR | 12.5% | 9.3% | .24 |
| DR, all pts | 31.3% | 36.5% | .16 |
| DR, axillary pCR | 20.3% | 15.9% | .25 |
| DR, no axillary pCR | 40.4% | 43.4% | .68 |
Authors read the benefit as carrying over the LRR and DM reductions seen in EORTC 22922 and MA.20 after adjuvant chemo into the NAT setting. They contrast prior HER2+ series (Danish 82b/c, SweBCG91-RT) suggesting relative radioresistance, and the ALTTO subset showing no RNI effect.
RNI pts had more adverse baseline features, so adjustment does heavy lifting. 'No RNI' may include unrecorded high tangents. Half the HER2+ cohort predated trastuzumab and was excluded from MVA, shrinking the subgroup behind the HR 0.237 estimate.
Supports RNI after NAT when nodal disease persists. Does not settle omission after axillary pCR; the authors defer that to B-51.
Retrospective, discretionary RNI with marked baseline imbalance; benefit appears only after covariate adjustment, univariate null. HER2+ trastuzumab DR signal is a subgroup estimate.
- Is RNI omission safe after cN1 to ypN0 conversion? active Standard or Comprehensive Radiation Therapy in Treating Patients With Early-Stage Breast Cancer Previously Treated With Chemotherapy and Surgery Phase NAn=1636 · primary completion 2023-09 · randomised regional nodal XRT vs breast/chestwall alonen=100 · primary completion 2028-06 · RNI de-escalation in biopsy-proven cN1 → pN0 after NATrecruiting Axillary Management in Breast Cancer Patients With Needle Biopsy Proven Nodal Metastases After Neoadjuvant Chemotherapy Phase NAn=1900 · primary completion 2030-02 · omits ALND + axillary RT in cN1 → ypN0 on SLNB, DFS 1° EP
- Does the HER2+ RNI benefit persist with modern HER2-directed NAT?
- Does the nomogram validate in an SLNB-staged cohort?
📚 Sources · 📄 1 paper
The longer read
The central tension in this paper is that the crude data and the adjusted data tell different stories. On the cumulative incidence curves, RNI did nothing visible: 10-year LRR 9.4% without and 8.1% with, and DR actually ran higher in the RNI group. Only after adjustment does RNI emerge as protective, halving LRR risk and trimming DR. That pattern is exactly what confounding by indication predicts. Physicians gave RNI to younger women with LVSI, higher clinical T and N category, and less response in breast and axilla, so the RNI group started out at higher risk. Adjustment for measured covariates is the right move, but it can only correct for what was recorded, and in a discretionary series the unrecorded reasons a radiation oncologist chose to treat the nodes (extent of residual disease beyond ypN category, imaging findings, extranodal extension) are the ones most likely to matter. A reader should treat the adjusted HRs as a plausible direction with an uncertain magnitude, not as an estimate of what RNI would achieve if randomly assigned.
The comparison group is also less clean than its label suggests. Two thirds of the patients without RNI received no radiation at all, so 'no RNI' here partly means 'no RT', and some LRR benefit may reflect breast or chest wall treatment rather than nodal coverage. In the other direction, tangent geometry was unavailable, so some 'no RNI' patients may have had incidental low axillary dose through high tangents. Those two biases push in opposite directions, and the paper cannot separate them.
The subgroup with the clearest clinical relevance, residual nodal disease after NAT, is also where the result is least surprising. An adjusted LRR HR of 0.476 in patients without axillary pCR fits what most radiation oncologists already do, and it is consistent with the authors' reading of EORTC 22922 and MA.20 as transferable to the neoadjuvant setting. The genuinely open question, whether nodes can be left alone after cN1 converts to ypN0, is the one this cohort cannot answer: 5.5% versus 4.6% LRR in pCR patients rests on too few events to model, and the authors explicitly defer to NSABP B-51/RTOG 1304.
The HER2+ trastuzumab DR signal (HR 0.237) is the most striking number and the most fragile. It comes from 145 patients, runs against prior series suggesting HER2+ disease gains less from RT, and predates dual HER2 blockade and post-neoadjuvant escalation for residual disease. It is hypothesis-generating at best.
The absence of any DSS or OS effect at 10 years is coherent with the randomized adjuvant trials and should temper any reading of the DR result as a survival argument. Practically, the nomogram is the more durable contribution: it lets a reader estimate absolute 10-year LRR from margins, grade, subtype and nodal response, which is the number that actually drives a coverage decision.