The Absolute Effect / Evidence / NSABP B-39 / RTOG 0413
NSABP B-39 / RTOG 0413
After lumpectomy for stage 0–II breast cancer, does accelerated partial-breast irradiation provide local control equivalent to whole-breast irradiation?
Registry: NCT00103181 · One underlying randomized study; aliases and follow-up reports consolidated.
Design, population and eligibility
Phase III randomized equivalence trial; 4,216 patients randomized (2,107 APBI; 2,109 whole breast).
Stage 0, I or II breast cancer; 61% postmenopausal, 81% hormone-receptor-positive; 24% DCIS, 65% node-negative invasive and 10% pN1.
Lumpectomy with histologically negative margins and 0–3 positive axillary nodes; broad eligibility compared with later selected APBI studies.
Median follow-up 10.2 years.
Treatment and comparator
Partial-breast irradiation: brachytherapy 34 Gy or external-beam 3D conformal radiotherapy 38.5 Gy in 10 fractions, twice daily over five treatment days.
Whole-breast irradiation 50 Gy in 25 fractions, with or without tumour-bed boost.
Primary endpoint: Ipsilateral breast tumour recurrence; equivalence margin required upper 90% CI of HR below 1.50.
Other endpoints: Recurrence-free interval, distant disease-free interval, DFS, OS and toxicity.
Key efficacy results
Ten-year IBTR-free rate was 95.2% with APBI versus 96.1% with whole-breast irradiation; corresponding recurrence risks 4.8% versus 3.9%. HR 1.22 (90% CI 0.94–1.58).
The prespecified equivalence criterion was not met because the upper confidence limit exceeded 1.50.
Ten-year recurrence-free interval was 91.8% versus 93.4%; HR 1.32 (95% CI 1.04–1.68), P=0.02. Distant disease-free interval and OS did not differ significantly.
Grade 3 toxicity was 9.6% with APBI versus 7.1% with whole breast; grade 4–5 toxicity 0.5% versus 0.3%.
Source-reported outcome visualization
Ten-year ipsilateral breast tumour recurrence-free rate · 10 years
Source-reported 10-year recurrence-free estimates. The absolute difference is small, but equivalence was not established.
Absolute effect
The 10-year IBTR risk was approximately 4.8% with APBI and 3.9% with whole-breast irradiation, an absolute difference of 0.9 percentage points. No NNT/NNH is assigned because the equivalence criterion was not met and the comparison was not a superiority trial.
NNT: Not assigned: failure to demonstrate equivalence does not establish a definitive superiority effect or a clinically stable NNT.
NNH: Not assigned; grade 3–5 toxicity differences were small and not converted to NNH without a prespecified significant harm endpoint.
Toxicity, harms and treatment burden
Grade 3 toxicity: 9.6% APBI versus 7.1% whole breast; grade 4–5: 0.5% versus 0.3%.
The APBI arm combined brachytherapy and external-beam techniques; treatment heterogeneity complicates toxicity and cosmesis interpretation.
The study's broad eligibility and twice-daily external-beam regimen limit direct transfer to current, selected APBI practice.
Quality of life and patient-reported outcomes
The primary publication did not establish a clear global QoL advantage for APBI. Treatment convenience must be balanced against local control, toxicity and cosmetic outcomes.
Subgroups, maturity and limitations
The population included DCIS and invasive disease, node-negative and limited node-positive disease, and multiple APBI techniques.
A post hoc impression that selected low-risk patients may do similarly cannot replace the failed overall equivalence test.
Ten-year follow-up is mature for local recurrence; OS and distant outcomes were not significantly different.
Clinical interpretation
Current-practice context · 1 November 2026 (full guideline freshness review pending): Modern APBI use is based on selected clinicopathological groups and technique-specific evidence. Avoid extrapolating this broad trial to every current APBI schedule or using its results to claim universal interchangeability.
Primary sources and registry
- Vicini et al. Lancet 2019 (primary report)
- AACR/SABCS primary results abstract
- ClinicalTrials.gov NCT00103181
Source hierarchy: primary randomized report and prespecified follow-up first. Missing numerical toxicity, discontinuation or QoL data are labelled rather than imputed.