The Absolute Effect / Evidence / Breast Cancer
EORTC 22922/10925
Clinical question: Does elective internal mammary and medial supraclavicular nodal irradiation improve long-term outcomes in stage I–III breast cancer?
Registry: NCT00002851 · One underlying study; aliases and related publications consolidated.
Phase III · randomized · Evidence cut-off: 8 November 2026
Design, population and eligibility
International randomized trial of elective internal mammary and medial supraclavicular nodal irradiation after surgery and standard systemic therapy.
4,004 patients with stage I–III breast cancer randomized between 1996 and 2004.
Axillary node-positive disease and/or centrally or medially located primary tumour, regardless of nodal status.
Median follow-up 22.2 years; final 20-year analysis published 2026.
Treatment and comparator
Intervention: Postoperative internal mammary and medial supraclavicular irradiation, 50 Gy in 25 fractions.
Comparator: No elective IM-MS nodal irradiation; other local and systemic treatments per protocol and era.
Primary endpoint: Overall survival; secondary endpoints included DFS, DMFS and breast cancer mortality.
Key outcomes
20-year OS: 61.8% control vs 61.0% IM-MS RT HR 1.00; p=0.967.
20-year DFS: 49.0% vs 48.2% HR 0.97; p=0.515.
20-year DMFS: 59.8% vs 58.9% HR 0.97; p=0.578.
20-year breast cancer mortality: 22.4% vs 18.6% HR 0.82 (95% CI 0.72–0.95); p=0.006.
Non-breast/unknown-cause mortality: 15.8% vs 20.4% HR 1.26 (95% CI 1.09–1.46); p=0.002.
At 20 years, IM-MS irradiation reduced breast cancer mortality by 3.8 percentage points but did not improve OS, DFS or DMFS. The late increase in non-breast/unknown-cause mortality offset the breast-cancer mortality reduction.
Absolute outcome visualization
Twenty-year breast cancer mortality, overall survival and competing mortality.
Values are reported at the stated time horizon; no outcome is derived from median survival. Single-arm data are not comparative effects.
NNT / NNH
The 3.8-percentage-point reduction in 20-year breast cancer mortality corresponds to a crude NNT of approximately 27 to prevent one breast cancer death by 20 years (1/0.038). This is a time-specific absolute-risk calculation from cumulative incidence; competing mortality and historical RT techniques must be considered. There is no OS NNT because OS did not improve.
Any cardiac disease was reported in 15.2% with IM-MS RT versus 11.7% without RT, an absolute excess of 3.5 percentage points (crude NNH ≈29 at 20 years; descriptive late morbidity). Lung fibrosis was 3.2% control vs 6.3% RT and cardiac fibrosis 1.7% vs 2.7%; crude excesses correspond to NNH 33 and 100 respectively, but these late-event comparisons are not a prespecified grade 3–5 NNH and should be treated descriptively.
Toxicity, grade 3–5 harms and treatment burden
The 20-year report recorded lung fibrosis 6.3% with IM-MS RT vs 3.2% control, cardiac fibrosis 2.7% vs 1.7%, and any cardiac disease 15.2% vs 11.7% (IM-MS RT vs control). Severe cardiac morbidity scores 3–4 were 1.9% vs 1.7%; severe lung morbidity 0.3% vs 0.0%.
25 fractions of nodal irradiation over five weeks in the historical protocol. Contemporary deep-inspiration breath hold, IMRT and proton techniques were not the treatment era tested in the original randomization.
Quality of life
The 20-year analysis focuses on survival, cause-specific mortality and late morbidity. No long-term QoL benefit is established.
Subgroups and applicability
Eligible patients had involved axillary nodes and/or central/medial tumours. The trial included stage I–III disease; absolute benefit may vary by baseline nodal and tumour-location risk.
Limitations
Treatment era 1996–2004; older RT planning and systemic therapies; long follow-up with competing risks and loss to follow-up; no OS benefit. Cause-of-death patterns are central to interpretation.
Clinical interpretation
IM-MS irradiation reduced long-term breast cancer mortality but did not improve 20-year OS, DFS or DMFS. Late non-breast mortality and cardiopulmonary effects temper the result. Contemporary heart/lung sparing may alter the harm profile but does not retrospectively change the trial's OS result.
Contemporary-practice note (8 November 2026): Historical trial results are distinct from current practice. Comprehensive guideline-level freshness review remains pending for this batch; current guidance and MDT context should inform treatment selection.