Locoregional recurrence after neoadjuvant chemotherapy: pooled analysis of nine trials
ForPrimary breast cancer after anthracycline/taxane NACT, incl. cT4a-d
TL;DR5-yr LRR 4.1% with pCR vs 9.5% without (multivariate HR 3.33); TNBC non-pCR 18.5%. Surgery type not independent after NACT.
Residual disease interacts with subtype: TNBC non-pCR LRR was 18.5% at 5y versus 4.2% after pCR, and HR-/HER2+ 14.8% vs 3.1%. With RT use unrecorded, these rates show where post-neoadjuvant RNI/PMRT escalation trials should focus. Within pCR, only cN+ trended toward risk (HR 1.61, p=0.054), relevant to node-based de-escalation.
In TNBC or HR-/HER2+ breast cancer with residual disease after NACT, this backs a high-LRR-risk read when weighing PMRT and nodal coverage; after pCR it shows low LRR across subtypes but cannot say whether omitting RT is safe, because RT delivery was not recorded.
Residual-disease TNBC (18.5%) and HR-/HER2+ (14.8%) carry the 5-yr LRR burden, while pCR flattens every subtype to 3.1-4.8%. RT delivery was unrecorded, so this sets risk strata for PMRT/RNI decisions after NACT rather than proving that RT can be omitted after pCR.
Mastectomy's crude LRR excess (11.3% vs 7.8% for BCS) disappears on multivariate analysis, and surgery type predicted LRR in no subtype. After NACT, choosing mastectomy over BCS did not independently improve local control; response and biology did.
13 details 4 trials watching
Exploratory individual-patient-data pooled analysis of 9 prospective GBG/AGO-B neoadjuvant trials (GeparDuo through GeparSepto, PREPARE, TECHNO), 1998-2013. N=10,075; median follow-up 67 mo. Pathologic response was analysed with landmarking (multivariate set N=9552).
Histologically confirmed primary invasive breast cancer, tumour mostly ≥2 cm; locally advanced cT4a-c and inflammatory cT4d eligible in all trials except GeparDuo. 63.6% HR-positive, 28.1% HER2-positive; pCR (ypT0 ypN0) in 20.4%.
Anthracycline/taxane-based NACT in all trials. Anti-HER2 therapy varied by trial (trastuzumab, lapatinib, pertuzumab in later trials; none in GeparDuo, GeparTrio, PREPARE). Endocrine therapy ≥5 years for ER/PgR-positive pts.
Protocols recommended adjuvant RT per national guidelines, with RNI indicated mainly by nodal stage. No patient-level RT data were available, so receipt, fields and dose are unknown.
LRR as first event (breast/chest wall, axillary, infra/supraclavicular nodes) measured from surgery; distant recurrence, second malignancy and death were competing events.
5-yr LRR 7.8% BCS vs 11.3% mastectomy, but surgery type was not an independent predictor in multivariate analysis. Age, cN, grade, subtype and pCR were independent; non-pCR HR 3.33 and TNBC HR 2.72.
| Stratum | pCR | Non-pCR |
|---|---|---|
| HR+/HER2- | 3.9% | 5.9% |
| HR+/HER2+ | 4.8% | 8.1% |
| HR-/HER2+ | 3.1% | 14.8% |
| TNBC | 4.2% | 18.5% |
| BCS | 4.0% | 8.8% |
| Mastectomy | 3.9% | 11.3% |
The authors place this alongside the EBCTCG meta-analysis, which found higher local recurrence after NACT than after adjuvant chemotherapy, especially with suboptimal local therapy. The NSABP B-51 question (RNI after nodal pCR) is the RT decision these data frame but cannot answer.
Anti-HER2 therapy was inconsistent across eras, which the authors use to explain the lower LRR in HR+/HER2+ disease. gBRCA status incomplete; the univariate surgery effect is likely confounded by selection.
Biology and response drive LRR after NACT more than surgery type. The non-pCR TNBC and HR-/HER2+ groups are where locoregional escalation has the most room to show benefit. With RT unrecorded, the low post-pCR rates are not evidence that RT can be omitted.
Large mature IPD pool of prospective trials confirming known LRR predictors after NACT. Exploratory, prognostic only, and blind to RT delivery, so it cannot test an RT decision.
- Is RNI or PMRT safely omitted after nodal pCR in initially cN+ pts? 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 · randomized regional nodal XRT vs none after chemo + surgeryn=100 · primary completion 2028-06 · RNI de-escalation in biopsy-proven cN1 → ypN0 after NATrecruiting Axillary Management in Breast Cancer Patients With Needle Biopsy Proven Nodal Metastases After Neoadjuvant Chemotherapy Phase NAn=1900 · primary completion 2030-02 · omit ALND + axillary RT if SLN negative post-NACT, cN1
- Does locoregional RT escalation reduce LRR in non-pCR TNBC? recruiting Internal Mammary Lymph Nodes Irradiation in High-risk Breast Cancer After Neoadjuvant Chemotherapy Phase 3n=722 · primary completion 2032-07 · phase 3 IMN RT vs none, ypN+ after NACT, DFS endpoint
- Do post-neoadjuvant T-DM1 or capecitabine lower LRR?
📚 Sources · 📄 1 paper
The longer read
The core message of this pooled analysis is that after neoadjuvant chemotherapy, locoregional risk follows tumour biology and response more than the operation. A crude gap between mastectomy and breast conservation (11.3% vs 7.8% at five years) vanishes once age, nodal stage, grade, subtype and pCR enter the model. That fits the authors' reading that mastectomy was given to higher-risk patients, and it should stop anyone from treating mastectomy after NACT as a local-control choice in itself. The decision that matters is what follows surgery.
For a radiation oncologist, the most useful output is the subtype-by-response grid. Without pCR, TNBC reached 18.5% and HR-/HER2+ 14.8% LRR at five years. With pCR, every subtype fell between 3.1% and 4.8%. In other words, achieving pCR erases most of the subtype gradient, and the high-risk tail sits almost entirely in residual-disease, hormone-receptor-negative tumours. Among pCR patients, only clinical node positivity hinted at residual risk (HR 1.61, p=0.054). That borderline signal is the part of the dataset most relevant to the current de-escalation debate. NSABP B-51 asked whether regional nodal irradiation can be dropped after nodal conversion, and this analysis neither supports nor undermines that question. It only shows that the baseline-node-positive, pCR population is not obviously risk-free.
The major caveat sets the ceiling on everything above: no patient-level radiotherapy data exist. Protocols recommended RT per guidelines, so most of these rates are presumably rates after some RT, but which patients got PMRT, which got nodal fields, and to what extent, is unknown. The low post-pCR LRR therefore cannot be read as the natural history of an untreated chest wall or nodal basin. It describes the outcome of a guideline-era mix of local therapy. The same problem cuts the other way for the high-risk groups. A TNBC non-pCR rate of 18.5% after presumably frequent RT points to radioresistance or systemic failure appearing locally first, not to an unirradiated population that RT would obviously rescue.
The treatment era limits generalisability. Several trials gave HER2-positive patients no anti-HER2 therapy, and none used post-neoadjuvant escalation with T-DM1 or capecitabine. Both have since improved outcomes in residual disease, though, as the authors note, their effect on LRR has not been reported. Contemporary non-pCR TNBC and HER2-positive patients probably carry lower locoregional risk than reported here, by an unknown margin. The HR+/HER2+ estimates are the most era-contaminated.
What would make this analysis misleading? A systematic link between RT omission and the high-risk strata, or the reverse, would distort any implied RT effect, and the dataset cannot test for either. It is best used as a risk-stratification map for designing and enrolling trials in post-neoadjuvant locoregional therapy, not as evidence for any individual RT decision.