Tumour bed boost after BCS+WBRT (Dutch cohort)
ForPost-BCS invasive breast cancer receiving WBRT, boost decision pending
TL;DR10yr IBTR 1.2% with 0-2 risk factors regardless of boost, supporting boost omission in the modern systemic era.
The decision this moves is boost omission, and the number that moves it is 10yr IBTR 1.2% in the 0-2 risk-factor group whether or not a boost was given, on 15,085 vs 13,845 pts. Note the ≥3 group ran higher WITH boost (3.3% vs 2.7%), which is allocation bias, not boost harm. Boost dose and fractionation are not in the source.
In a post-BCS patient over 40 with grade 1-2, hormone-receptor-positive disease receiving guideline-concordant systemic therapy, this supports omitting the tumour bed boost; it does not resolve the boost question for pts carrying three or more risk factors.
The omission decision rests on 10yr IBTR of 1.2% in the 0-2 risk-factor group with and without boost (15,085 vs 13,845 pts). The ≥3 stratum ran higher WITH boost (3.3% vs 2.7%), a signature of risk-based allocation rather than boost harm. Boost dose and fractionation are absent from the source.
| Risk factors | N no boost | N boost | 5yr no boost | 5yr boost | 10yr no boost | 10yr boost |
|---|---|---|---|---|---|---|
| 0-2 | 15,085 | 13,845 | 0.6% | 0.7% | 1.2% | 1.2% |
| ≥ 3 | 149 | 733 | 1.3% | 2.9% | 2.7% | 3.3% |
| Uncertain | 592 | 944 | 0.8% | 3.3% | 1.4% | 3.6% |
+2 more figures
9 details
Population-based Dutch cohort from the Netherlands Cancer Registry linked to pathology, on behalf of the DBRT group. Treatment years 2012-2016, follow-up to 10 years. Observational, no randomisation and no adjusted comparison reported in source.
Breast-conserving treatment with or without an RT boost, N=31,348 across the three risk strata. Stratification is by a count of five risk factors: age ≤40, grade 3, triple-negative, guideline-indicated systemic therapy not adequately given, and no pCR after neoadjuvant chemo in TNBC or HER2+.
Whole-breast RT with or without a tumour bed boost. Boost dose, fractionation, technique (photon vs electron vs SIB) and the WBRT schedule are not reported in the source slides, which limits transfer to a specific departmental protocol.
Primary: ipsilateral breast tumour recurrence (IBTR), histologically confirmed, identified by an algorithm over pathology report codes and free text. Reported as cumulative incidence at 5 and 10 years by risk-factor count. Benchmarked against the Assisi thresholds: omission acceptable at <3% 10yr IBTR with boost, <6% without.
IBTR was low in every stratum. The only cell crossing an Assisi threshold was ≥3 risk factors treated with a boost at 10 years, and even there the no-boost value in the same stratum was lower.
EORTC 22881-10882 established that a boost roughly halves IBTR, and that trial's control-arm event rates were an order of magnitude above these. IMPORT HIGH and the 2024 Assisi think tank both moved the field toward de-escalating or restricting the boost; this cohort supplies the contemporary absolute rates those recommendations assumed but could not show.
Boost was allocated by guideline-based risk, so the boost groups are adversely selected and the raw contrast understates any boost effect; the higher rate in the ≥3 boost group is the visible signature of that confounding. The ≥3 no-boost cell holds only 149 pts, and the 'uncertain' stratum (592 / 944) shows a boost-no-boost gap wide enough to suggest unmeasured risk is driving allocation there too.
The finding is about absolute rather than relative benefit: a preserved 50% relative reduction applied to a 1.2% 10-year event rate is not worth five extra fractions and a fibrosis penalty. What the cohort cannot say is whether the boost is the reason those low-risk rates are low, since roughly half the low-risk group received one.
Registry cohort, no randomisation and no adjusted effect estimate; boost allocation confounded by risk. Supports the direction already set by IMPORT HIGH and Assisi thresholds.
- Which ≥3 risk-factor subgroups actually benefit from a boost
- Whether boost omission holds under randomised testing in low-risk pts
- Boost dose and technique used across this cohort
📚 Sources · 🐦 1 tweet
Day TWO of #ESTRO26 Coverage by OncoAlert 🚨
— OncoAlert (@OncoAlert) May 16, 2026
Is a boost to the tumour bed still indicated after breast-conserving surgery and whole-breast radiotherapy in the era of modern systemic therapy? Presented by Femke Froklage 🇳🇱 #RadOnc ☢️
We aimed to identify a subgroup of breast… pic.twitter.com/RqK5r9XPqW
The longer read
The argument here is entirely about absolute risk, and it is worth being explicit that this cohort does not challenge the boost's relative efficacy at all. EORTC 22881-10882 remains the reference: a boost cuts ipsilateral recurrence by roughly half, and nothing in a Dutch registry series can overturn a 5,000-patient randomised trial with 20-year follow-up on that point. What has changed is the denominator. In the EORTC era the 10-year local recurrence risk after whole-breast RT alone sat in the range where halving it delivered several absolute percentage points. In this 2012-2016 cohort, 10-year IBTR in the 0-2 risk-factor group is 1.2%. A relative halving of 1.2% is a fraction of a percent, spread across a population large enough that the number needed to boost runs into the hundreds. That is the whole case, and it is a strong one, because it does not require the boost to have stopped working.
The methodological weakness runs in a specific and legible direction. Boost allocation followed Dutch guideline risk criteria, which means the pts who got a boost were, on average, the pts at higher baseline recurrence risk. This is not subtle confounding to be adjusted away; it is the visible structure of the data. The ≥3 risk-factor stratum recorded a higher 10-year IBTR with boost (3.3%) than without (2.7%), and the uncertain stratum shows the same inversion more starkly (3.6% vs 1.4%). Read naively that says boost causes recurrence, which nobody believes. Read correctly it says allocation was doing exactly what it was designed to do, and it calibrates how much of the apparent equivalence in the 0-2 group might be a boost effect being cancelled by adverse selection. That direction of bias, importantly, works against the paper's conclusion rather than for it: if the boosted pts were higher risk and still landed at the same 1.2%, the true omission penalty is at most small, possibly zero, but the cohort cannot bound it.
What should move a reader's confidence upward is the size and the ascertainment. Nearly 29,000 pts in the low-risk strata with pathology-report-level recurrence capture is a far better estimate of the contemporary event rate than any trial's control arm, and the contemporary event rate is the quantity that actually decides this. What should move confidence downward is everything about the ≥3 group, where 149 unboosted pts cannot support a subgroup conclusion, and the authors are appropriately careful to say the question remains open there rather than manufacturing a recommendation.
The practical gap is that the source gives no boost dose, fractionation, or technique. That matters less for an omission decision than it would for a dose question, but it means this cohort cannot be read as evidence about a simultaneous integrated boost versus a sequential one, and it cannot be mapped onto a department that boosts to 16 Gy versus one that uses 10 Gy. For a reader deciding whether to keep boosting a 55-year-old with a grade 2 node-negative tumour on adjuvant endocrine therapy, that is a decision this supports leaving behind. For the young, triple-negative, or residual-disease-after-neoadjuvant patient, it changes nothing, and the honest reading is that those are precisely the pts a randomised de-escalation trial would still need to enrol.