Nomogram for locoregional recurrence in T1-2N1 breast cancer (PMRT decision-making)
ForT1-2N1 breast, mastectomy + ALND, no neoadjuvant therapy
TL;DRHigh-risk nomogram tier (score >18): PMRT 5y LRR 2.2% vs 14.9%, HR 0.164; no LRR association in low or intermediate tiers.
The omission case is the low-risk tier: 5y LRR 0.4% without PMRT and 0/160 vs 2/252 events, so any RT gain there is bounded near zero. Benefit concentrates above score 18 (14.9% → 2.2% LRR, 7.0% BCM reduction) with chest wall plus supraclavicular/infraclavicular fields at 46-50 Gy, i.e. standard comprehensive PMRT.
In T1-2N1 pts after mastectomy and ALND without neoadjuvant therapy, ER-positive, grade I-II, LVI-negative T1 disease with 1-2 nodes, this supports a PMRT omission discussion; it does not address pts after neoadjuvant therapy.
Low-risk tier baseline 5y LRR is 0.4% with 0/160 vs 2/252 events, bounding any PMRT gain near zero and supporting omission. The high-risk tier (score >18) carries the benefit, 14.9% → 2.2% LRR, with standard chest wall plus supraclavicular/infraclavicular fields at 46-50 Gy.
13 details 3 trials watching
Retrospective single-institution cohort of 1141 consecutive pts treated 2001-2012. Nomogram from Fine and Gray competing-risks regression in 623 unirradiated pts; risk tiers by decision tree (CHAID); internal bootstrap validation only. Median follow-up 74.9 mo.
T1-2, 1-3 positive axillary nodes, mastectomy with negative margin plus ALND (median 17 nodes dissected). Excluded neoadjuvant therapy, synchronous metastases, other malignancy. All received adjuvant chemotherapy, 90.5% anthracycline- or taxane-based; median age 47.
46-50 Gy in 23-25 fx, CT-based planning. Fields: ipsilateral chest wall, infraclavicular and supraclavicular; IMN only for central tumors, axilla only with ECE.
LRR as first event (chest wall or regional nodes), DR and BCM, all as cumulative incidence with death as a competing risk.
Nomogram C index 0.73. Risk tiers by score: low ≤10 (n=412), intermediate (n=385), high >18 (n=344). PMRT associated with lower LRR, DR and BCM only in the high-risk tier.
| Group | LRR | DR | BCM |
|---|---|---|---|
| All pts (n=1141) | 1.6% vs 6.0%, HR 0.248 (0.121-0.509), P<.001 | HR 0.608 (0.425-0.871), P=.006 | P=.130 |
| Low risk (n=412) | P=.930 | P=.220 | P=.390 |
| Intermediate risk (n=385) | P=.240 | P=.500 | P=.430 |
| High risk (n=344) | 2.2% vs 14.9%, HR 0.164 (0.067-0.403), P<.001 | 8.4% vs 17.6%, HR 0.491 (0.288-0.837), P=.009 | 6.3% vs 13.3%, HR 0.371 (0.191-0.719), P=.003 |
EBCTCG showed PMRT benefit in 1-3 node pts, but its control 10-yr LRR of 21% is well above this cohort's 6.0% 5-yr unirradiated rate, which is the gap a risk model tries to exploit. SUPREMO (NCT00966888) is the prospective test of selective PMRT in this setting.
PMRT recipients skewed to adverse features, so within-tier comparisons carry residual confounding. Era of low trastuzumab use (26.2%) and no capture of nodal size or margin width; a prior 242-pt nomogram (Wadasadawala) used different factors.
The model mostly formalizes features clinicians already weigh (T stage, node count, ER, grade, LVI) into a score. Its value is the low-risk tier's very low baseline LRR, which supports omission, but the benefit estimates need external validation before they guide individual decisions.
Retrospective, clinician-allocated PMRT; risk cutoffs data-derived and tested on the same cohort without external validation. Low-risk null rests on 2 events.
- External validation in an independent, contemporary cohort
- PMRT benefit in the intermediate-risk tier with adequate events n=1106 · primary completion 2026-12 · phase 3 PMRT vs no PMRT in pN1, 7y DFS endpoint
- Role of PMRT after neoadjuvant therapy active Standard or Comprehensive Radiation Therapy in Treating Patients With Early-Stage Breast Cancer Previously Treated With Chemotherapy and Surgery Phase NAn=1636 · primary completion 2023-09 · randomised chestwall/nodal XRT after chemo + surgeryn=100 · primary completion 2028-06 · RNI de-escalation in cN1 → ypN0 after neoadjuvant tx
📚 Sources · 📄 1 paper
The longer read
The useful output of this paper is a prognostic statement, not a treatment-effect estimate. The nomogram is trained on unirradiated pts, so what it quantifies is baseline locoregional risk after mastectomy and ALND, and on that question it is reasonably credible: a C index of 0.73 from five factors that any breast radiation oncologist already weighs. The decision tree then cuts that score into tiers, and the low tier carries a 5-year LRR of 0.4% without PMRT. Whatever the true relative effect of RT, the absolute gain available there is close to nothing, and that holds regardless of whether the PMRT comparison in the paper is confounded.
The treatment-effect half is weaker. PMRT was allocated by clinicians, and the paper itself reports that recipients were younger, larger, more node-positive and higher grade. Adjustment for measured covariates cannot reach what was not recorded, and the authors concede they lacked nodal size and margin width, both of which plausibly drove referral. The high-risk result (LRR 14.9% to 2.2%, BCM 13.3% to 6.3%) is directionally consistent with EBCTCG, which the authors cite as translating a 16.5% local-control gain into a 7.9% breast cancer mortality gain at 20 years. A skeptic would note that here a breast cancer mortality gain of 7.0% is already visible by 5 years, earlier than a pure local-control mechanism would predict, which is the pattern one expects when the treated and untreated groups differ in ways the model did not capture.
The null results in the low and intermediate tiers should be read as underpowered rather than negative. Two LRR events in 252 unirradiated low-risk pts cannot show a benefit, and the authors say as much. That matters most for the intermediate tier, where baseline 5-year LRR is 6.1%: that is not trivial, and this cohort cannot tell a reader whether RT helps there. That tier is where the real disagreement in T1-2N1 PMRT lives.
Transfer to current practice runs in one direction. Every pt here had ALND with a median of 17 nodes removed, adjuvant chemotherapy, no neoadjuvant therapy, and HER2-directed therapy in only a quarter of HER2-positive cases. Modern systemic therapy should push baseline risk down further, which makes the low tier's omission case stronger and the high tier's absolute benefit likely smaller than reported. Pts treated after neoadjuvant therapy are outside the data entirely, as the authors note in pointing to NSABP B-51.
The model would be wrong in a way that matters if external validation shows the high-risk tier was an artifact of fitting cutoffs to this cohort's events, or if SUPREMO finds no meaningful benefit across the 1-3 node population. Until then it is a reasonable way to structure an omission conversation for favorable T1-2N1 disease, not a basis for withholding RT from the intermediate tier.