Breast Cancer and Its Surgical Treatment: An Evidence-Based Review (Part 1)

Evidence Grading Key

Recommendation strength follows GRADE: Strong or Conditional/Weak.

A note on single-arm trials. Several of the most exciting results in surgical de-escalation (Kuerer's surgery-omission cohort, LUMINA) are single-arm. A single-arm trial with a spectacular result is Level IV design carrying Level II persuasive weight when the observed event rate is far below any plausible historical control. This review labels them by design and comments on the effect size separately, rather than letting enthusiasm inflate the grade.

Three organising claims of this review:

1. Breast cancer surgery has been progressively de-escalated for fifty years without loss of survival, because local therapy controls local disease while systemic therapy determines survival. Radical mastectomy → modified radical → breast conservation → sentinel node → omission of sentinel node is one continuous arc. (Level I)
2. Axillary surgery has largely completed its transition from a therapeutic to a staging procedure, and is now losing even that role in low-risk disease, because systemic therapy decisions are increasingly made on tumour biology and genomic assays rather than nodal count. (Level I–II)
3. Omission of breast surgery itself is the current frontier and is not yet standard of care anywhere. The single-arm data are striking; randomised confirmation does not exist. Presenting it as an option outside a trial would be a serious misreading of the evidence. [EXP]

Executive Summary

What is established

What is emerging but not yet standard

The central intellectual shift

Breast cancer surgery is moving from an anatomy-driven discipline (how much tissue must be removed to control this tumour?) to a biology- and response-driven one (what does this tumour's behaviour and its response to systemic therapy tell us about how much local treatment is still needed?).

Two facts underlie this:

1. Local therapy does not usually determine survival. Halsted's hypothesis — that breast cancer spreads centrifugally and can be cured by wider local excision — was falsified by NSABP B-06. Distant outcome is set by micrometastatic biology at diagnosis and by systemic therapy.
2. Response to neoadjuvant therapy is itself a biomarker. A patient whose cN1 axilla converts to ypN0 has demonstrated chemosensitivity, and that demonstration carries prognostic information that anatomical staging cannot. NSABP B-51 is the cleanest expression of this idea: the response, not the original stage, determined whether further local therapy helped.

Where the de-escalation logic must stop

Honest counterweight to the above:

• De-escalation trials are dominated by postmenopausal HR+/HER2− disease. In INSEMA, the great majority of patients were postmenopausal with HR+/HER2− tumours. Applying its conclusion to a 38-year-old with a 3 cm triple-negative tumour is unsupported.
• Non-inferiority margins are choices, not facts. A trial that declares non-inferiority with a margin of HR 1.27 has accepted a definable amount of possible harm as tolerable. Whether the patient in front of you would accept that trade is a separate question.
• Follow-up duration matters differently by subtype. HR+ disease recurs late — beyond 10, 15, even 20 years. A 5-year non-inferiority result in a luminal population is reassuring but not conclusive.
• Omitting staging removes information. If the sentinel node is not examined, nodal status cannot inform adjuvant decisions. This is acceptable when the systemic plan would not change — which is why SLNB omission is best supported precisely where genomic assays and tumour biology already determine treatment.
• Access and quality are prerequisites. Omitting axillary surgery depends on high-quality axillary ultrasound. Omitting breast surgery would depend on expert multi-modality imaging and standardised vacuum-assisted biopsy. In settings without these, de-escalation is not de-escalation; it is under-staging.

Methodology

Approach

PRISMA-informed rather than PRISMA-compliant. Sources identified through targeted retrieval of: current guideline documents (NCCN Breast Cancer, ESMO, ASCO/ASTRO/SSO consensus statements, ASBrS); landmark randomised trials defining each domain; recent systematic reviews and meta-analyses (2018–2026); and trials reported after the most recent guideline revisions.

Source hierarchy

1. Adequately powered low-risk-of-bias RCTs with mature follow-up
2. Meta-analyses restricted to randomised evidence (EBCTCG analyses given particular weight)
3. Current guideline recommendations
4. Meta-analyses of observational data with explicit certainty grading
5. Large registry/population cohorts
6. Single-arm prospective trials and institutional series

Biases explicitly tracked

• Surgical trials cannot be blinded. Patient-reported outcomes — a major endpoint in de-escalation trials — are particularly susceptible.
• Selection bias in the de-escalation literature. Trials that enrol low-risk patients will find low event rates in both arms; this makes non-inferiority easier to demonstrate and simultaneously limits generalisability. This is not a flaw, but it is a constraint that is frequently dropped when results are summarised.
• Non-adherence. COMET had approximately 29% crossover from assigned arm. Intention-to-treat and per-protocol analyses gave meaningfully different pictures (2-year ipsilateral invasive cancer 4.2% vs 5.9% ITT; 3.1% vs 8.7% per protocol). Both are reported here; neither alone is the answer.
• Industry sponsorship. Systemic therapy trials cited (KEYNOTE-522, KATHERINE, OlympiA, monarchE, DESTINY-Breast programme) are industry-sponsored. This does not invalidate them but is relevant to endpoint selection, comparator choice, and framing.
• Publication bias is likely to inflate apparent benefit of novel reconstructive and robotic techniques, where negative single-centre experiences are rarely published.
• Length-time and lead-time bias dominate the DCIS and screening literature and are addressed directly in a later part of this series.

Causal language

Observational associations are reported as associations. Causal language is reserved for randomised evidence. Pathological complete response is not treated as equivalent to cure at the individual level — it is a strong prognostic marker whose surrogacy for survival varies by subtype and is weakest in HR+/HER2− disease.

Epidemiology

Burden

Breast cancer is the most commonly diagnosed cancer in women worldwide and a leading cause of cancer death in women. Incidence correlates with Human Development Index; mortality-to-incidence ratios are markedly worse in low- and middle-income settings, where late-stage presentation and limited access to systemic therapy, radiotherapy, and reconstructive surgery drive outcome gaps far larger than any surgical technique difference discussed in this review.

This disparity deserves emphasis. The entire de-escalation literature presupposes reliable access to high-quality imaging, pathology, radiotherapy, and systemic therapy. In settings lacking radiotherapy access, mastectomy is not an outdated operation — it is the correct operation, because BCS without radiotherapy carries substantially higher local recurrence.

Risk factors

Surgical relevance of germline status: BRCA1/2, PALB2, TP53, PTEN, and CDH1 carriers face elevated ipsilateral and contralateral risk that may justify bilateral mastectomy — a decision distinct from treating the index cancer. CDH1 carriers specifically develop lobular cancer, which is mammographically occult and multifocal, complicating conservation. Germline testing should be resolved before definitive surgery whenever the result could change the operation.

Surgical Anatomy

Structural

The breast is a modified apocrine gland lying on the pectoralis major fascia between roughly the 2nd and 6th ribs, from sternal edge to mid-axillary line, with the axillary tail of Spence extending into the axilla. It comprises 15–20 lobes draining via lactiferous ducts to the nipple, embedded in stroma and fat, supported by Cooper's suspensory ligaments running from deep fascia to dermis. Tumour tethering of these ligaments produces skin dimpling.

The retromammary space between posterior breast fascia and pectoral fascia is the plane of mastectomy dissection.

The subcutaneous plane is the anterior limit. Its correct identification is the central technical problem of skin- and nipple-sparing mastectomy: dissecting too superficially devascularises the flap and causes necrosis; dissecting too deeply leaves breast tissue and residual cancer risk. There is no universally agreed flap thickness — this is a real and under-acknowledged source of variability in nipple-sparing mastectomy outcomes.

Vascular

Lymphatic drainage

Approximately 75% of breast lymph drains to the axilla, the remainder chiefly to internal mammary nodes, with minor interpectoral (Rotter's) and supraclavicular routes. Drainage is predominantly to a first-echelon node or small group — the anatomical basis of sentinel node biopsy.

Axillary levels (defined by pectoralis minor)

Standard ALND = levels I and II, typically yielding ~10 or more nodes. Level III dissection is reserved for gross disease.

Internal mammary nodes

Drain medial and central tumours. Positive in a minority of patients. Not routinely biopsied; when clinically relevant, addressed by regional nodal irradiation rather than surgery. This is one of the oldest de-escalation decisions in the field — extended radical mastectomy including internal mammary node dissection was abandoned because it added morbidity without survival benefit.

Nerves at risk in the axilla

Preservation of the intercostobrachial nerve where oncologically safe is a low-cost intervention with meaningful quality-of-life return.

Molecular Classification — Why It Now Precedes the Operation

Surgical planning in 2026 begins with the core biopsy immunohistochemistry, not with the tumour's dimensions.

HER2-low and HER2-ultralow (IHC 1+, or 2+/ISH−, and faint incomplete staining ≤10%) are not biological subtypes but predictive categories created by the efficacy of trastuzumab deruxtecan in that expression range. They do not currently change the operation; they change systemic options in advanced disease.

Why this table drives surgery

Three concrete consequences:

1. pCR probability determines whether neoadjuvant therapy is worth pursuing for surgical de-escalation. Offering neoadjuvant chemotherapy to a luminal A tumour to shrink it for conservation frequently fails and delays definitive treatment. Offering it to a HER2+ or triple-negative tumour frequently succeeds.
2. Residual disease after neoadjuvant therapy is now actionable. KATHERINE established that residual invasive disease after neoadjuvant HER2-directed therapy should be treated with T-DM1 rather than continued trastuzumab, with a large improvement in invasive disease-free survival. This means the surgical specimen is a therapeutic decision instrument, not merely a record. It is a strong argument against omitting surgery outside trials: without the specimen, the escalation decision cannot be made on the same evidence base.
3. Genomic assays have displaced nodal count for chemotherapy decisions in much HR+/HER2− disease (TAILORx for node-negative, RxPONDER for 1–3 nodes, MINDACT for clinical-high/genomic-low). This is the deep reason axillary staging is losing its role: the axilla was staged in order to decide about chemotherapy, and that decision is now increasingly made from the tumour itself.

Actionable alterations (surgical relevance in brackets)

Staging: Anatomic and Prognostic

TNM (AJCC 8th edition, abbreviated)

The prognostic stage — the most important staging change of the last decade

AJCC 8 introduced prognostic stage, which incorporates grade, ER, PR, HER2, and (in defined situations) the 21-gene recurrence score alongside anatomic T, N, M.

Why this matters surgically: two tumours with identical anatomy can differ by two or more prognostic stage groups. A 2.5 cm grade 1 ER+/PR+/HER2− node-negative tumour (anatomic stage IIA) may downstage substantially on prognostic criteria; a 1.5 cm grade 3 triple-negative node-negative tumour upstages. Prognostic stage encodes the same insight that drives the whole de-escalation programme: biology outweighs size.

Stage-specific surgical implications

Principles of Surgical Treatment

Goals, in order

1. Complete removal of the index cancer with clear margins (local control)
2. Accurate assessment of what remains (staging/therapeutic information for adjuvant decisions)
3. The best achievable cosmetic and functional outcome consistent with (1) and (2)
4. Alignment with the patient's own priorities, which are legitimate determinants and not tie-breakers

Margins — the SSO/ASTRO/ASCO consensus

(Level I for invasive; ⊕⊕⊕⊝ for DCIS.)

This consensus was a de-escalation. Before it, re-excision rates in some series exceeded 20–30%; adoption of "no ink on tumour" substantially reduced reoperation without increasing recurrence. It is the clearest example in breast surgery of evidence directly removing an operation from practice.

Localisation of non-palpable lesions

Evidence position: non-wire techniques consistently improve logistics and patient experience; evidence for reduced positive margin rates is inconsistent. Meta-analyses have variously shown modest margin improvement or none. The honest summary is that these are workflow and patient-experience innovations with an unproven oncologic dividend. (⊕⊕⊝⊝)

Shared decision-making — a substantive requirement, not a formality

Breast surgery is unusual in oncology because more than one option is frequently oncologically correct. When BCS + RT and mastectomy offer equivalent survival, the choice legitimately belongs to the patient. This has two demanding implications:

• The surgeon must present equivalence honestly, including that mastectomy does not improve survival and does not eliminate local recurrence risk.
• The surgeon must not present de-escalation as automatically better. Some patients, fully informed, choose mastectomy for reasons of anxiety, surveillance burden, radiotherapy avoidance, or symmetry preference. That is a legitimate choice, not a failure of counselling.

The known asymmetry: rates of contralateral prophylactic mastectomy in average-risk women have risen substantially in several health systems despite absent survival benefit — evidence that counselling frequently fails in the direction of over-treatment driven by risk perception rather than risk.

Breast-Conserving Surgery versus Mastectomy

The foundational evidence

NSABP B-06 randomised women with tumours ≤4 cm to total mastectomy, lumpectomy alone, or lumpectomy plus breast irradiation. At 20 years there was no significant difference in overall survival between the three arms; radiotherapy after lumpectomy substantially reduced ipsilateral breast tumour recurrence. Multiple concurrent European trials (Milan, EORTC, Danish) reached the same conclusion.

(Level I, ⊕⊕⊕⊕. This is among the most robust findings in surgical oncology.)

EBCTCG meta-analysis established the quantitative relationship underlying modern radiotherapy decisions: radiotherapy after BCS reduces 10-year recurrence risk substantially, and roughly one breast cancer death is avoided by year 15 for every four recurrences prevented by year 10. This "4:1 rule" is why local control is not irrelevant to survival even though the primary determinant is systemic.

The observational signal that BCS may be better

Multiple large population-based analyses since 2015 have reported superior survival with BCS + RT compared with mastectomy. Taken at face value this would be startling.

It should not be taken at face value. The most plausible explanations are confounding by indication (patients selected for mastectomy have larger, more advanced, or more multifocal disease, or comorbidity precluding radiotherapy), the survival contribution of radiotherapy itself, and unmeasured differences in access and adjuvant therapy delivery. Randomised evidence shows equivalence; observational evidence showing superiority almost certainly reflects selection.

The correct summary is equivalence. (⊕⊕⊕⊕ for equivalence; ⊕⊝⊝⊝ for superiority.)

Comparison table

Indications for mastectomy

Oncoplastic surgery — the technique that expanded conservation

Oncoplastic BCS combines tumour excision with volume displacement (local tissue rearrangement, therapeutic mammoplasty) or volume replacement (local perforator flaps — LICAP, TDAP, AICAP). It permits larger resections while preserving shape.

Evidence position: consistently improves cosmetic outcome and permits conservation in patients who would otherwise require mastectomy; oncologic equivalence is supported by cohort data rather than randomised trials. (⊕⊕⊝⊝) Practical caveats: higher complication rates than simple lumpectomy, potential delay to adjuvant therapy if complications occur, and greater difficulty in radiological surveillance and re-excision if margins are positive because tissue has been rearranged. Careful clip marking of the tumour bed is mandatory.

Mastectomy variants

Nipple-sparing mastectomy is oncologically acceptable in selected patients: tumour not involving the nipple–areola complex, adequate tumour–nipple distance, no nipple discharge or Paget's, and technically achievable flap thickness. Retroareolar margin assessment is standard. Local recurrence in the retained NAC is low in reported series. The evidence base is cohort-derived, not randomised (⊕⊕⊝⊝), and outcomes are strongly operator-dependent — NAC necrosis rates vary widely between centres and are the principal complication.

Axillary Management and the De-escalation Programme

This is the domain where breast surgery has changed most, and the central subject of this review. It is treated at length.

The historical logic and why it collapsed

ALND was originally performed for three reasons: regional control, staging, and presumed therapeutic benefit. Each has been progressively undermined.

• The therapeutic rationale was undermined by NSABP B-04, which found no survival difference between radical mastectomy, total mastectomy with axillary radiotherapy, and total mastectomy with delayed axillary surgery for clinical recurrence.
• The staging rationale is being undermined by genomic assays that determine chemotherapy from the tumour rather than the axilla.
• The regional control rationale survives, but radiotherapy achieves it with far less morbidity (AMAROS).

What remains is morbidity: lymphedema affecting a substantial minority of ALND patients, shoulder dysfunction, chronic pain, and intercostobrachial sensory disturbance — all lifelong, all in patients who are largely cured.

Sentinel lymph node biopsy

Technique. Radioisotope (technetium-99m), blue dye, indocyanine green fluorescence, or superparamagnetic iron oxide, alone or in combination. Dual-tracer techniques give the highest identification rates and lowest false-negative rates and remain the reference standard where available. ICG fluorescence has grown rapidly because it avoids nuclear medicine logistics; its performance appears comparable in reported series (⊕⊕⊝⊝).

Established position: SLNB is the standard axillary staging procedure for clinically node-negative invasive breast cancer, with equivalent survival and dramatically lower morbidity than ALND. (Level I, Strong)

The completion-ALND question: can positive sentinel nodes be left alone?

Why SENOMAC matters more than its predecessors

Z0011's conclusions were persistently criticised on three grounds: it enrolled only BCS patients receiving whole-breast tangential radiotherapy (which incidentally treats part of the low axilla), it closed early with fewer events than planned, and it excluded T3 tumours.

SENOMAC addresses all three. It included T1–T3 tumours, included a large mastectomy cohort, and was powered on overall survival rather than a composite. Its conclusion — that omission of completion ALND should be standard for 1–2 sentinel macrometastases — therefore rests on a substantially broader base.

One important caveat repeatedly emphasised by the investigators: in SENOMAC, adjuvant treatment followed local standard of care, and a majority of patients received nodal radiotherapy. The trial establishes that the surgery can be omitted. It does not establish that both the surgery and the nodal radiotherapy can be omitted. Conflating these would over-extend the result.

(Level I, ⊕⊕⊕⊕, Strong: omission of completion ALND for 1–2 sentinel macrometastases in the studied population.)

The frontier: omitting sentinel node biopsy entirely

Reading these trials correctly

The eligibility criteria are the finding. INSEMA required:

• Clinically and ultrasonographically node-negative axilla
• Breast-conserving surgery
• Whole-breast radiotherapy (which treats a portion of the low axilla)
• Predominantly postmenopausal, HR+/HER2− disease

Omitting SLNB in a patient who does not meet these — a mastectomy patient, a patient not receiving radiotherapy, a young patient with triple-negative disease, a patient whose axilla was not assessed by quality ultrasound — is not supported by these data.

The trade-off is also real, not free: axillary recurrence was higher in the omission arm (1.0% vs 0.3%), an absolute difference of 0.7 percentage points that did not translate into a survival difference over the observed follow-up. In HR+ disease with a long recurrence tail, longer follow-up remains genuinely informative.

(Level II, ⊕⊕⊕⊝, Conditional: SLNB may be omitted in patients meeting SOUND/INSEMA-like criteria after explicit discussion of the trade-off.)

Axillary management after neoadjuvant systemic therapy

This is the most technically demanding area in current axillary surgery.

The problem: in patients who present cN+ and convert to clinically node-negative after neoadjuvant therapy, plain SLNB has an unacceptably high false-negative rate (historically reported above 10% in ACOSOG Z1071 and SENTINA when standard technique was used).

The solutions, and the evidence for each:

Current position: TAD has been widely adopted for cN1 patients converting to clinical node-negativity, with ALND reserved for residual nodal disease (ypN+) or where the marked node cannot be retrieved. Long-term randomised oncologic outcome data for TAD are still maturing — adoption has outpaced definitive evidence, which is defensible given the morbidity avoided but should be stated plainly.

NSABP B-51/RTOG 1304 — response as the determinant of radiotherapy

Design: cT1–T3, N1, M0 patients with biopsy-proven nodal disease who achieved ypN0 after neoadjuvant chemotherapy were randomised to regional nodal irradiation or no regional nodal irradiation (after mastectomy: chest wall + RNI vs no RT; after BCS: whole-breast RT + RNI vs whole-breast RT alone). NCT01872975. Published NEJM 2025.

Result: the addition of regional nodal irradiation did not reduce invasive breast cancer recurrence or breast cancer death. The primary endpoint (invasive breast cancer recurrence-free interval) showed no benefit.

Significance: this is the purest demonstration in breast oncology that treatment response is a better guide to local therapy than initial anatomic stage. A patient who presents cN1 and converts to ypN0 has different biology from a patient who presents cN1 and remains ypN+, and treating them identically over-treats the responder.

Limitations, stated honestly: median follow-up is not yet long for a population that includes HR+ disease; the trial used conventional fractionation (50 Gy/25) whereas hypofractionation is now standard; and subgroup behaviour by molecular subtype requires longer observation. About a fifth of enrolled patients had characteristics that placed them at higher residual risk.

(Level II, ⊕⊕⊕⊝, Conditional-to-Strong: regional nodal irradiation may be omitted in ypN0 patients meeting B-51 criteria.)

Consolidated axillary decision summary

Can Surgery Be Omitted Altogether?

This question is asked repeatedly and it deserves an unambiguous answer in two parts.

DCIS: active monitoring

COMET (NCT02926911, JAMA 2025). 995 women aged ≥40 with newly diagnosed grade 1–2, hormone receptor-positive, HER2-negative DCIS without invasive disease, randomised to guideline-concordant care (surgery ± radiotherapy) or active monitoring (6-monthly ipsilateral mammography, biopsy for imaging change, surgery on invasive diagnosis). Endocrine therapy permitted in both arms.

Median follow-up 36.9 months. Invasive tumour characteristics did not differ significantly between groups. A parallel quality-of-life analysis found comparable physical, emotional, and psychological outcomes.

Interpretation — with the necessary caution:

• The result is short-term. DCIS has a natural history measured in decades; two years is a screening interval, not an outcome horizon.
• 29% of participants did not adhere to their assigned arm, which is why the ITT and per-protocol analyses diverge so markedly. The ITT result is the primary and appropriate analysis; the per-protocol divergence is at least partly explained by invasive disease being found at surgery in the treated arm (i.e. it was already there, not caused by treatment).
• Companion trials — LORIS (UK, closed early with 188 patients), LORD (Europe), LORETTA/JCOG1505 (Japan, single-arm, tamoxifen alone) — will contribute, but the field is under-powered relative to the question's importance.
• This is a genuine overdiagnosis problem, not a treatment-efficacy problem. DCIS incidence rose steeply with mammographic screening without a proportionate fall in invasive cancer incidence — the signature of detecting lesions that would not have progressed. Active monitoring addresses that, and is therefore conceptually important beyond its effect size.

(Level II design, ⊕⊕⊝⊝ overall given follow-up duration. Conditional: reasonable to offer to carefully selected low-risk DCIS patients within trials or with rigorous informed consent and structured surveillance. Not yet a general standard.)

Invasive cancer: omission of breast surgery after pCR

The rationale. With modern neoadjuvant regimens, pCR rates approach or exceed 60% in triple-negative and HER2-positive disease. If a patient has no residual invasive cancer, the argument runs, the operation removes nothing.

The obstacle. Imaging cannot reliably confirm pCR. Mammography, ultrasound, and MRI all misclassify residual disease in both directions. Radiological complete response is not pathological complete response.

The MD Anderson programme (Kuerer et al.). A prospective non-randomised trial enrolled patients with cT1–2 N0–1 triple-negative or HER2-positive unicentric disease whose lesions shrank to <2 cm on imaging after neoadjuvant systemic therapy. Image-guided vacuum-assisted core biopsy of the tumour bed (minimum 12 cores) determined pCR. Patients with biopsy-determined pCR proceeded to whole-breast radiotherapy with boost and no breast surgery; those with residual disease had standard surgery. Of the initial 50 patients, 31 (62%) had biopsy-determined pCR.

Preplanned 5-year outcomes (JAMA Oncology 2025;11(5):529–534): no recurrences observed. Companion work reported a vacuum-assisted biopsy false-negative rate of 3.2% (95% CI 0.1%–16.7%).

Why this is not yet practice-changing, stated plainly:

1. It is a single-arm, non-randomised trial in a small, extraordinarily selected cohort. Unicentric disease, <2 cm residual on imaging, favourable subtypes, expert breast imaging, and a standardised 12-core sampling protocol. The eligibility criteria exclude the majority of patients.
2. The confidence interval on the false-negative rate is wide — the upper bound of 16.7% is not reassuring, and reflects a small denominator.
3. The specimen carries therapeutic information that is lost if surgery is omitted. KATHERINE established that residual invasive disease after neoadjuvant HER2-directed therapy mandates escalation to T-DM1. If no specimen exists, that decision cannot be made on the evidence base it was built on. A vacuum-assisted biopsy showing no residual disease is not equivalent to a whole-specimen pathological assessment.
4. Radiotherapy is not omitted in these protocols. This is de-escalation of surgery, not of local therapy.
5. Larger confirmatory studies are ongoing, including expanded MD Anderson cohorts and international trials (Korean, European, Russian, and Brazilian protocols are registered). None has reported randomised outcomes.

(Level IV design with a striking effect size. ⊕⊝⊝⊝ / [EXP]. Omission of breast surgery for invasive cancer is investigational and should occur only within a clinical trial.)

The honest answer to "can breast cancer eventually be treated without surgery?"

Plausibly, in a minority, eventually. The requirements for that to become safe are specifiable:

Until these are met, the correct framing for patients is: this is a promising research direction, not an available option.

Selected Landmark Trials

Key Points

1. BCS + radiotherapy and mastectomy give equivalent survival. Population data suggesting BCS is superior reflect selection, not effect.
2. The adequate margin is no ink on tumour for invasive cancer and 2 mm for DCIS with radiotherapy. Wider is not better.
3. Completion ALND is not required for 1–2 sentinel node macrometastases. SENOMAC extends this to T1–T3 and to mastectomy patients, with overall survival as the endpoint.
4. SLNB itself can be omitted in patients meeting SOUND/INSEMA criteria — but the criteria are the finding, and most enrolled patients were postmenopausal with HR+/HER2− disease undergoing BCS with whole-breast radiotherapy.
5. After neoadjuvant chemotherapy converts cN1 to ypN0, regional nodal irradiation adds nothing (NSABP B-51). Response beats initial stage as a guide to local therapy.
6. TAD is the standard sentinel-based approach for cN1 patients converting after neoadjuvant therapy; ALND is reserved for ypN+.
7. Active monitoring for low-risk DCIS was non-inferior at 2 years in COMET. Two years is short for a disease measured in decades. Reasonable within trials or with rigorous consent; not a general standard.
8. Omission of breast surgery for invasive cancer is investigational. The 5-year single-arm data are striking and the selection criteria are narrow. Trials only.
9. The surgical specimen is now a therapeutic decision instrument (KATHERINE). This is the strongest under-discussed argument against surgery omission outside trials.
10. Contralateral prophylactic mastectomy does not improve survival in average-risk women. Rising rates reflect a counselling failure in the direction of over-treatment.
11. Every de-escalation result above presupposes access to quality imaging, pathology, radiotherapy, and systemic therapy. Where those are absent, de-escalation becomes under-treatment.
12. All statements describe populations, not patients. Individual management belongs to a multidisciplinary team.

Abbreviations

• ADM, acellular dermal matrix
• ALND, axillary lymph node dissection
• BCS, breast-conserving surgery
• BCSS, breast cancer-specific survival
• cCR, clinical complete response
• CPM, contralateral prophylactic mastectomy
• ctDNA, circulating tumour DNA
• DCIS, ductal carcinoma in situ
• DIEP, deep inferior epigastric perforator
• DMFS, distant metastasis-free survival
• EBCTCG, Early Breast Cancer Trialists' Collaborative Group
• ER, oestrogen receptor
• ET, endocrine therapy
• iDFS, invasive disease-free survival
• ICG, indocyanine green
• LCIS, lobular carcinoma in situ
• MRD, molecular residual disease
• NAC, nipple–areola complex (context: also neoadjuvant chemotherapy — disambiguated in text)
• NST, neoadjuvant systemic therapy
• pCR, pathological complete response
• PR, progesterone receptor
• RNI, regional nodal irradiation
• RS, recurrence score
• SERD, selective oestrogen receptor degrader
• SLNB, sentinel lymph node biopsy
• TAD, targeted axillary dissection
• T-DM1, trastuzumab emtansine
• T-DXd, trastuzumab deruxtecan
• TNBC, triple-negative breast cancer
• VAB/VACB, vacuum-assisted (core) biopsy
• WBRT, whole-breast radiotherapy
• ypN0, pathologically node-negative after neoadjuvant therapy

1. COMET Study Investigators, Hwang, E. S., Hyslop, T., Lynch, T., Ryser, M. D., Weiss, A., et al. (2025). Active monitoring with or without endocrine therapy for low-risk ductal carcinoma in situ: The COMET randomized clinical trial. JAMA, 333(11), 972–980. https://doi.org/10.1001/jama.2024.26698
2. de Boniface, J., Filtenborg Tvedskov, T., Rydén, L., Szulkin, R., Appelgren, M., Reimer, T., et al. (2026). Omission of completion axillary dissection in patients with breast cancer and sentinel lymph node macrometastases: Overall survival and patient-reported arm morbidity from the randomized SENOMAC trial [Abstract LBA503]. Journal of Clinical Oncology, 44(17_suppl).
3. Gentilini, O. D., Botteri, E., Sangalli, C., Galimberti, V., Porpiglia, M., Agresti, R., et al. (2023). Sentinel lymph node biopsy vs no axillary surgery in patients with small breast cancer and negative results on ultrasonography of axillary lymph nodes: The SOUND randomized clinical trial. JAMA Oncology, 9(11), 1557–1564. https://doi.org/10.1001/jamaoncol.2023.3759
4. Kuerer, H. M., Smith, B. D., Diego, E. J., Johnson, H. M., Shaitelman, S. F., Mitchell, M. P., Boughey, J. C., White, R. L., Hunt, K. K., Yang, W. T., & Rauch, G. M. (2025). Selective elimination of breast surgery for invasive breast cancer: A nonrandomized clinical trial. JAMA Oncology, 11(5), 529–534.
5. Mamounas, E. P., Bandos, H., White, J. R., Julian, T. B., Khan, A. J., Shaitelman, S. F., et al. (2025). Omitting regional nodal irradiation after response to neoadjuvant chemotherapy. New England Journal of Medicine. https://doi.org/10.1056/NEJMoa2414859
6. Morrow, M. (2025). Sentinel-lymph-node biopsy in early-stage breast cancer — Is it obsolete? New England Journal of Medicine, 392(11), 1134–1136. https://doi.org/10.1056/NEJMe2414899
7. Reimer, T., Stachs, A., Veselinovic, K., Kühn, T., Heil, J., Polata, S., et al. (2025). Axillary surgery in breast cancer — Primary results of the INSEMA trial. New England Journal of Medicine, 392(11), 1051–1064. https://doi.org/10.1056/NEJMoa2412063
8. Whelan, T. J., Smith, S., Parpia, S., Fyles, A. W., Bane, A., Liu, F.-F., et al. (2023). Omitting radiotherapy after breast-conserving surgery in luminal A breast cancer. New England Journal of Medicine, 389(7), 612–619. https://doi.org/10.1056/NEJMoa2302344
9. National Comprehensive Cancer Network. (2026). NCCN clinical practice guidelines in oncology: Breast cancer. NCCN. (Verify the current version number at nccn.org before citation; NCCN revises breast cancer guidance several times per year.)