IMPORT HIGH
ForInvasive early breast, pT1-3 pN0-pN3a M0, post-BCS, requiring tumour bed boost
3.7% vs 3.5% 10yr IBTR (48Gy SIB vs 40+16Gy)
95% CI 2.6-5.3 vs 2.4-5.0; 53Gy/15F 5.5% (4.1, 7.3)
TL;DR10yr IBTR 3.7% with 48Gy/15F SIB vs 3.5% with 40Gy/15F + 16Gy/8F sequential; 53Gy/15F higher at 5.5%.
The decision this hardens is delivery, not dose: 48Gy/15F SIB holds at 3.7% (2.6, 5.3) IBTR at 10 years against 3.5% (2.4, 5.0) for a sequential 16Gy/8F phase, in a higher risk group. 53Gy/15F sits at 5.5% (4.1, 7.3), so escalation buys nothing.
For a woman after breast conserving surgery for pT1-3 pN0-pN3a invasive disease who needs a tumour bed boost, the 10-year data support the integrated 48Gy/15F arm over a separate sequential boost; they do not speak to boost omission or to 5-fraction whole-breast schedules.
Three weeks, one plan: the boost is integrated into 15 fractions with a modest dose reduction to whole breast distant from tumour, and IBTR at 10 years matches the sequential 16Gy/8F phase. Escalating the integrated boost to 53Gy/15F does not improve local control.
| Dose group | 10yr IBTR (95% CI) | 10yr OS abs. diff vs 40Gy/15F |
|---|---|---|
| 40Gy/15F + 16Gy/8F | 3.5% (2.4, 5.0) | reference |
| 48Gy/15F (3.2Gy/F) | 3.7% (2.6, 5.3) | -0.5 (-3.0, 2.8) |
| 53Gy/15F (3.5Gy/F) | 5.5% (4.1, 7.3) | 1.5 (-1.4, 5.1) |
+1 more figure
8 details 4 trials watching
Three-arm randomised multicentre trial, 1:1:1, N=2617 across 76 UK hospitals, recruited 2009-2015. Annual clinical follow-up to 10 years; PRO and photographic assessment collected only to 5 years.
Women ≥18 after breast conserving surgery for invasive early breast cancer, pT1-3, pN0-pN3a, M0, all requiring a tumour bed boost. Described as a higher-than-average risk group.
40Gy/15F + 16Gy/8F sequential boost (N=871), 48Gy/15F SIB at 3.2Gy/F (N=874), 53Gy/15F SIB at 3.5Gy/F (N=872). The SIB arms escalate to the regions at highest risk with a modest dose reduction to whole breast distant from tumour, all delivered in 3 weeks.
Endpoint reported here: ipsilateral breast tumour relapse at 10 years. The original sample size calculation assumed a 5% control rate at 5 years. Absolute OS difference and clinician-assessed normal tissue effects also reported.
The 5-year ordering holds at 10 years: the two lower-dose groups sit close together and 53Gy/15F stays highest. Both absolute OS differences vs 40Gy/15F have intervals containing zero.
Moderate/marked effects at 10 years were given as bounds across all randomised groups: <18% breast distortion or shrinkage, <10% induration, <2% telangiectasia, <2% breast oedema. No per-arm split reported in source.
EORTC 22881-10882 established the tumour bed boost itself; IMPORT HIGH asks how to deliver it and whether more dose helps. At 10 years, integration works and escalation does not, the same ranking the 5-year publication (Coles et al. Lancet 2023;401:2124-37) reported.
PRO and photographic assessment stopped at 5 years, so the 10-year toxicity comparison rests on clinician scoring reported as all-group bounds, not per-arm rates. Observed IBTR also ran below the 5% control rate the sample size calculation assumed.
The practical read is fraction count, not dose: a boost folded into 15 fractions removes the separate 16Gy/8F phase with no 10-year IBTR cost, while 53Gy/15F returns nothing. What the trial does not settle is whether the same integration transfers to 5-fraction whole-breast schedules.
CONSORT flow
Mature 10yr follow-up of a 2617-pt randomised trial; extends the 5-year Lancet 2023 read rather than changing it. No formal 10yr non-inferiority margin stated in source.
- Simultaneous integrated boost within 5-fraction whole-breast schedules recruiting Ultra-hypofractioNated Adjuvant Radiotherapy ± sImultaneous Integrated Boost for Low-risk Breast Cancer Patients Phase 2n=65 · primary completion 2025-10 · ultra-hypofx WBI +/- SIB, low-risk, phase 2recruiting Ultra Hypo-fractionated Adjuvant Whole Breast Radiation Therapy With Simultaneous Integrated Boost for Early-Stage Breast Cancer (H-ASSIST) Phase 2n=90 · primary completion 2028-02 · 5-fraction WBI with SIB tumor bed boost, phase 2
- Patient-reported cosmesis beyond 5 years, unmeasured after photographic follow-up ended n=139 · primary completion 2025-12 · 10y registry with cosmesis + QoL assessmentsn=50 · primary completion 2028-09 · cosmesis + PROMs to 60mo after ultra-short WBI/SIB
📚 Sources · 🐦 1 tweet
Day THREE of #ESTRO26 Coverage by OncoAlert 🚨
— OncoAlert (@OncoAlert) May 17, 2026
Ten-year results of the IMPORT HIGH trial (ISRCTN47437448): Dose escalated simultaneous integrated boost radiotherapy in early breast cancer Presented by Charlotte Coles 🇬🇧 #RadOnc ☢️
Ten-year IMPORT HIGH trial data show that a… pic.twitter.com/7RqVy2SrQm
The longer read
The question IMPORT HIGH was built to answer was never whether a tumour bed boost helps. EORTC 22881-10882 settled that, and every pt in this trial received a boost. What was open was how to give it, and whether the ability to escalate boost dose without extending treatment would convert into fewer relapses. Ten years of follow-up answers both parts, and the two answers point in opposite directions.
On delivery, the result is about as good as this design can produce. IBTR at 10 years is 3.7% (2.6, 5.3) for 48Gy/15F simultaneous integrated boost against 3.5% (2.4, 5.0) for the sequential 40Gy/15F plus 16Gy/8F control, in a group selected for higher than average recurrence risk. A schedule finishing in three weeks matches one that runs a separate boost phase of eight further fractions. For a department that is throughput; for a pt it is eight fewer attendances, and at a decade neither costs local control.
On escalation, the answer is flat. 53Gy/15F sits at 5.5% (4.1, 7.3), highest of the three, with a lower bound of 4.1 above the control's point estimate of 3.5. Nothing here supports pushing the boost further, and the direction of the point estimate is the wrong one for a hypothesis that more dose to the highest-risk region should reduce relapse. The honest read is that the tumour bed is not dose limited in this population across these levels, so remaining gains in early breast RT sit in fractionation and volume, not intensity.
Two features should move confidence, in different directions. Event rates came in below plan: the sample size calculation assumed 5% in the control group at five years, and the observed 10-year figures in the two lower-dose groups fall under that. A trial powered on an event rate it did not see carries wider effective uncertainty than its N suggests, which bites hardest on the escalation comparison, where the question is whether 5.5% against 3.5% is real. Separately, PRO and photographic assessment stopped at five years. The 10-year toxicity data are clinician-scored and reported as bounds across all randomised groups, under 18% for breast distortion or shrinkage, under 10% for induration, under 2% for telangiectasia and for oedema, with no per-arm split in the source. The late toxicity claim for the 53Gy arm therefore rests on the coarsest instrument at the latest timepoint, which is exactly where a fibrosis signal would surface.
The overall survival differences, -0.5 (-3.0, 2.8) and 1.5 (-1.4, 5.1) against 40Gy/15F, are uninformative by design and should be read that way: both intervals contain zero and the trial was never sized for survival. What would have to be true for the delivery conclusion to be wrong is that the SIB arms carry a late cosmetic or fibrosis penalty that clinician scoring at 10 years missed. Truncated PRO follow-up leaves that open rather than answered, and it is the one thing a reader adopting 48Gy/15F should keep watching.