Fewer lymph nodes removed, same survival

Overview

Giuliano AE et al.. JAMA 2017 · PMID 28898379 · DOI 10.1001/jama.2017.11470

In women with 1-2 positive sentinel nodes, skipping full axillary dissection did not worsen 10-year survival.

Does the study support the specific claim? Partly · Methodological quality: high

Original publication

Plain-language summary

Scientific in-depth review

1. Full reference

Authors

Giuliano AE et al.

Journal

JAMA

Year

2017

Identifier

PMID 28898379 DOI 10.1001/jama.2017.11470

Verification

verified – PubMed efetch 2026-10-25

Retracted

no

Corrections

none

Extent checked

full text

Conflicts of interest and funding

NCI grants (U10CA180821, U10CA180882 and others) to the Alliance; NCI had a role in design, conduct, analysis and manuscript review, without veto right. Disclosures: institutional per-capita accrual payments from ACOSOG; one author speaker fees from Genentech and Genomic Health; statisticians compensated by ACOSOG.

2. Research question

Is sentinel node dissection alone non-inferior to axillary lymph node dissection for 10-year overall survival?

3. Study design and population

Study design

Randomised controlled trial – Phase 3 randomised non-inferiority trial (ACOSOG Z0011), 115 sites

Population

Women with clinical T1-T2 invasive breast cancer, no palpable adenopathy, 1-2 positive sentinel nodes, treated with lumpectomy, whole-breast irradiation and systemic therapy

Country

USA (115 academic and community sites; ACOSOG/Alliance, NCI-funded)

Period

Enrolment May 1999 to December 2004; database frozen 29 September 2015; median follow-up 9.3 years

Sample size

891 randomised; 35 withdrew consent; 856 analysed (ITT: 436 SLND alone, 420 ALND); planned 1900 patients and 500 deaths, closed early (110 deaths observed)

Inclusion and exclusion criteria

Adult women with histologically confirmed invasive breast cancer <=5 cm (clinical T1-T2), no palpable axillary adenopathy, 1-2 sentinel nodes with metastases detected by H&E; planned lumpectomy, tangential whole-breast irradiation and adjuvant systemic therapy; third-field irradiation prohibited. Not applicable to mastectomy, >2 positive nodes, palpable nodes, no whole-breast irradiation or neoadjuvant therapy (authors).

4. Methods and endpoints

Exposure / intervention

Sentinel lymph node dissection alone

Comparator

Axillary lymph node dissection

Endpoints

Overall survival (primary), disease-free survival, regional recurrence

Statistical methods

Stratified randomisation (age, hormone receptor status, tumour size); non-inferiority with HR margin 1.3 and 1-sided 95% CI from Cox regression; ITT and as-treated analyses (similar); Kaplan-Meier, log-rank; adjusted Cox models; exploratory hormone-receptor subgroup analysis; lost-to-follow-up censored. Final 10-year OS analysis not prespecified.

5. Actual results

Results

Nodes removed: median 2 (SLND) vs 17 (ALND); 27.3% of ALND patients had further macrometastases in non-sentinel nodes. Deaths 51 vs 59. 10-year OS 86.3% (82.2-89.5%) vs 83.6% (79.1-87.1%); HR 0.85 (1-sided 95% CI 0-1.16), non-inferiority P=0.02; adjusted HR 0.93 (1-sided 95% CI 0-1.28). 10-year DFS 80.2% (75.6-84.1%) vs 78.2% (73.5-82.2%), HR 0.85 (0.62-1.17), P=0.32. One nodal recurrence after year 5 in SLND group, none in ALND. Adjuvant systemic therapy 97.0% vs 96.0%; 18.9% received protocol-prohibited nodal irradiation (balanced).

Absolute risks

10-year overall survival 86.3% vs 83.6%; 10-year disease-free survival 80.2% vs 78.2%; deaths 51/436 vs 59/420

Authors' conclusion

For women with T1-T2 breast cancer, no palpable adenopathy and 1-2 positive sentinel nodes treated with breast-conserving therapy and systemic therapy, 10-year OS with SLND alone was non-inferior to ALND; routine ALND is not supported in this population. Findings should not be extrapolated to other settings (mastectomy, >2 nodes, palpable nodes, no whole-breast irradiation, neoadjuvant therapy).

6. Strengths

• Randomised multicentre phase 3 trial with 10-year follow-up and hard primary endpoint (overall survival)
• Stratified randomisation, balanced baseline characteristics
• ITT and as-treated analyses concordant
• Changed international practice, with consistent later trials

7. Weaknesses

• Closed early: 891 of 1900 planned patients and 110 of 500 planned deaths, so underpowered; wide non-inferiority margin (HR 1.3)
• Adjusted HR upper bound (1.28) close to the margin
• Open label; 18.9% received prohibited nodal-field irradiation (balanced between groups)
• 35 withdrew consent; losses to follow-up censored
• 10-year analysis not prespecified
• Results limited to breast-conserving therapy with whole-breast irradiation; not to mastectomy or neoadjuvant settings

8. Risk of bias

Orienting tool

RoB 2

Type of assessment

structured assessment based on the full text by a single reviewer (no formal double review)

Full formal assessment

no

Overall risk of bias

some concerns

Rationale

Overall unchanged. Main additional limitation is imprecision from early closure (110 of 500 planned deaths) and a generous non-inferiority margin. Simplified single-reviewer judgement from the full text; a formal RoB 2 assessment requires two independent reviewers and the protocol (available as supplement, not reviewed).

Full-text review (25 October 2026): Added country (USA), period details, inclusion criteria and stated scope limits, early closure (891/1900 patients, 110/500 deaths), ITT numbers, adjusted HR 0.93 (upper bound 1.28), CIs for 10-year survival, protocol deviations (18.9% nodal irradiation), non-prespecified 10-year analysis, funding/COI. RoB 2 overall unchanged (some_concerns) with domains.

9. Assessment of the statistical interpretation

Non-inferiority met for the unadjusted HR (upper 1-sided bound 1.16 < 1.3), but the trial was underpowered and the adjusted upper bound (1.28) is near the margin. Absolute 10-year survival was numerically higher without ALND (86.3% vs 83.6%).

10. Relevance to the specific claim

Supports that extensive lymph-node removal is unnecessary in selected breast-cancer patients; sentinel node biopsy (staging) and tumour removal remained.

11. Independent contrary findings

Applies only to the defined population.

12. Transferability to Germany and Europe

Basis of current European practice.

13. Overall methodological judgement

Practice-changing RCT showing that completion axillary dissection can be omitted in a defined group of women with limited sentinel-node involvement who receive breast-conserving surgery, radiotherapy and systemic therapy; it supports de-escalation within standard oncology, not avoidance of tumour surgery or staging.

14. Open questions

Generalisability to mastectomy patients.

15. Linked fact checks

• Surgical oncologists do not understand what tumours are: they remove the tumour and 'clean out' the lymph nodes, which he regards as wrong. – Studies with mixed or unclear results