EBCTCG 20-year recurrence after 5 years of endocrine therapy
ForER+ T1-T2, 0-9 nodes, disease-free after 5y adjuvant endocrine therapy
TL;DRAfter stopping endocrine therapy at 5y, distant recurrence in years 5-20 ran from 13% (T1N0) to 41% (T2N4–9), at a steady annual rate.
In ER+ T1-T2 early breast cancer disease-free at 5 years, this frames the extension discussion around TN status, with T1N0 still carrying 13% distant risk over years 5 to 20; it does not address genomic-assay-selected or trastuzumab-treated HER2+ pts.
11 details
EBCTCG individual patient data meta-analysis of 88 trials. Kaplan-Meier and Cox regression, stratified by trial and treatment, from year 5 to year 20.
62,923 women with ER-positive T1-T2 disease, 0-9 positive nodes, diagnosed before age 75, scheduled to stop endocrine therapy at 5 years and event-free at year 5. 63% tamoxifen, 17% AI, 20% a tamoxifen-AI sequence; all had lymph-node dissection.
Distant recurrence, any breast-cancer event (distant, locoregional, contralateral), and death from breast cancer by log-rank subtraction.
Recurrences continued at a steady rate from year 5 to 20 in every TN category. Given TN status, grade and Ki-67 carried only moderate independent value after year 5; PR and HER2 status were not predictive.
| Nodal status | T1 | T2 |
|---|---|---|
| N0 | 13% | 19% |
| N1–3 | 20% | 26% |
| N4–9 | 34% | 41% |
| Grade | Distant recurrence | Any recurrence or contralateral BC |
|---|---|---|
| Low | 10% | 17% |
| Moderate | 13% | 22% |
| High | 17% | 26% |
| Nodal status | Distant recurrence | Death from breast cancer |
|---|---|---|
| N0 | 22% | 15% |
| N1–3 | 31% | 28% |
| N4–9 | 52% | 49% |
The source restates EBCTCG data that 5y tamoxifen cut recurrence by about 30% in years 5-10 versus no therapy. Extension trials such as ATLAS and MA.17 test the intervention; this analysis supplies the untreated baseline risk that their proportional benefit acts on.
Grade and Ki-67 were often locally assessed, and Ki-67 was available in only 7692 pts. Few pts received trastuzumab (2%), and treatment eras predate modern staging and systemic therapy.
No TN subgroup had a late risk low enough to dismiss extension outright, even low-grade T1N0 at 10% distant recurrence. Whether extension actually lowers mortality, and in whom it outweighs toxicity, is left to randomised data.
Large multi-trial IPD prognostic analysis, not a randomised test of extension. Quantifies baseline late risk; extended-therapy benefit must come from separate randomised trials.
- Does a genomic low-risk score identify pts safe to stop at 5 years?
- Does extended endocrine therapy reduce breast-cancer mortality, not just recurrence?
- How much do modern-era treatments lower these late risks?
📚 Sources · 📄 1 paper
The longer read
The usable output of this analysis is a baseline, not a treatment effect. Extended endocrine therapy trials report proportional reductions, and the authors argue those reductions look roughly similar across risk groups, so the absolute gain from continuing past 5 years depends almost entirely on what a given woman's risk would be if she stopped. This paper supplies that denominator at a scale and follow-up no single trial can match: 62,923 women across 88 trials, followed to year 20. The headline finding is that the risk curve does not flatten. In every tumor-node category, distant recurrences kept accruing at a near-constant annual rate through the second decade, so the clinical intuition that ER+ disease that survives 5 years without event is largely cured does not hold.
The second finding is that anatomy outranks biology once the first 5 years have passed. Tumor size and especially nodal status remained powerful determinants of late distant recurrence, and their effects were roughly additive, running from 13% for T1N0 to 41% for T2N4–9. Grade and Ki-67, which carry strong prognostic weight early, lost most of their independent value after year 5, and PR and HER2 status lost theirs entirely. That pattern argues against using early-recurrence proliferation markers to talk a woman out of extension; the original TN stage remains the better guide. Even low-grade T1N0 disease carried 10% distant recurrence and 17% any breast-cancer event over years 5 to 20, which is the number that undercuts any claim of a group too low-risk to discuss.
Several design features should shape how confidently a reader applies these figures. The analysis is prognostic, drawn from within randomised trials but not testing extension itself, so it says nothing directly about how much continuing therapy would help. The cohort is old: about half were diagnosed before 2000, and better staging, surgery, radiation and systemic therapy since then likely shift absolute risks downward. Two biases pull in opposite directions. Risks were measured in women scheduled for 5 years regardless of adherence, so women who actually completed therapy probably did somewhat better, while the authors estimate that 5 to 10% of distant recurrences went unrecorded, so true rates are probably higher. HER2+ disease was largely untreated with trastuzumab, and no gene-expression data were available, which is the most important gap for current practice, where multigene assays already sort node-negative ER+ disease.
The decision this informs is the 5-year conversation in a disease-free woman. For node-positive or T2 disease, a steady late risk in the 20 to 40% range makes the absolute benefit of extension large enough to weigh seriously against toxicity. For T1N0, the risk is lower but not negligible, and the authors note that for women in reasonable health the competing risk of other-cause death will not erase it. What the paper does not settle is whether a genomic low-risk signature identifies a subgroup whose late risk truly falls below the threshold for benefit, and whether extension changes breast-cancer mortality rather than recurrence alone. For the radiation oncologist, the read is indirect: locoregional recurrence followed the same TN dependence as distant recurrence, so late locoregional events in the second decade are expected, not anomalous, in higher-stage ER+ disease.