Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jul 7.
Published in final edited form as: JAMA Surg. 2026 Sep 1;161(9):861–869. doi: 10.1001/jamasurg.2026.2340

Lumpectomy margins and local recurrence in DCIS: Results from NRG Oncology/NSABP B-35 Trial

Irene L Wapnir 1, Reena S Cecchini 2, James J Dignam 3, Stewart J Anderson 4, Jiahe Li 2, J Marie Suga 5, Adam M Brufsky 6, Judith O Hopkins 7, Kenneth Zeitzer 8, Janice A Lyons 9, Alison Conlin 10, Henry Mark Kuerer 11, Shahzad Siddique 12, Laura A Vallow 13, Kathy S Albain 14, Mary E Cianfrocca 15, Thomas B Julian 16, Jean-François Boileau 17, Eleftherios P Mamounas 18, Norman Wolmark 19
PMCID: PMC13324939  NIHMSID: NIHMS2176271  PMID: 42384406

Abstract

Importance

NRG Oncology/NSABP B-35 prospectively collected margin width data on post-menopausal women with HR-positive DCIS who underwent lumpectomy, whole breast irradiation (WBI), and randomly assigned adjuvant anastrozole or tamoxifen therapy. This permitted analysis of outcomes per margin width.

Objective

Local recurrence rates were similar in the two endocrine-assigned treatment arms, permitting analysis of margin width effect on ipsilateral breast tumor recurrence (IBTR).

Design/Setting/Participants

NSABP B-35 was a phase 3, double-blind trial in which 3,104 patients were randomized to either five years of tamoxifen or anastrozole. Post-menopausal women with HR-positive DCIS and tumor-free margins were eligible. Enrollment was from 1-6-2003-6-15-2006, in academic and community hospitals members of the NSABP. Lumpectomy margin width data was prospectively collected within three months of randomization. A pathology form classified margins as: positive (ink-on-tumor), close (<1mm), or negative (≥1mm). For the negative margin subgroup, closest margin width was stated separately. Thus, an ancillary analysis using 1mm (2,707 patients) and 2mm (2,546 patients) margin width partitions was performed.

Interventions

There were no specific interventions based on lumpectomy margin width.

Main Outcomes/Measures

In this analysis, IBTR was the most common first event, occurring in 90/2,707 (3.3%) patients: 24/502 (4.8%) with <1mm margin width and 66/2,205 (3.0%) with ≥1mm margin width. Ten-year unadjusted cumulative incidence of IBTR events was 5.6% vs 4.0% for <1mm vs ≥1mm (P=0.04) margins. Using 2mm as the discriminant threshold for margin width, 39/879 (4.4%) patients with <2mm margins and 49/1,667 (2.9%) with ≥2mm margins experienced an IBTR first. Ten-year unadjusted cumulative incidence of IBTR events with ≤2mm margins was 5.3% vs 3.8% for ≥2mm margins (P=0.05). In models adjusting for other patient and tumor factors, margin width was not a significant predictor of IBTR risk (2mm threshold HR=1.33 [95% CI=0.86–2.06]).

Conclusions/Relevance

Absolute differences in IBTR rates using < or ≥1mm and < or ≥2mm margin width groupings in post-menopausal women with HR-positive DCIS receiving lumpectomy, WBI, and adjuvant endocrine therapy, were small. Omission of re-excision lumpectomies based on margin widths of <1mm or <2mm can be reconsidered in appropriate patients.

Introduction

Margin width is a surgically modifiable factor that can mitigate breast cancer local recurrence after lumpectomy. In patients with DCIS, the occurrence of an ipsilateral breast tumor recurrence (IBTR) often leads to mastectomy, and if the recurrence is invasive, to additional systemic treatments as well as a small absolute increase in mortality.1 Some of the largest ductal carcinoma in situ (DCIS) trials, such as SweDCIS2 and NSABP B-24,3 did not require tumor-free margins, whereas NSABP B-174and EORTC5 simply defined margins as tumor-free. Yet other trials such as RTOG 9804,6 required at least 3 mm margins for eligibility. None of these randomized clinical trials addressed the relationship between specific margin widths and local recurrence. Moreover, many published studies grouped close margins with involved/positive margin cases, thus blurring the interpretation of what may be considered an adequate negative margin.7

The current 2mm margin width guideline for DCIS is based on the recommendations of The Consensus Guideline On Margin Width for Breast-conserving Surgery, which were derived from the systematic meta-analysis of 20 studies involving 7,883 patients who underwent lumpectomy and whole breast irradiation (WBI).8 Their analysis revealed the statistically significantly lower risk of IBTR with 2 mm vs 0 or 1 mm margins (OR 0.51, 95% CI 0.31–0.85, P=0.01). Furthermore, margins ≥2 mm were not associated with lower local recurrence rates.

The NRG Oncology/NSABP B-35 study is the largest randomized trial to date that prospectively collected margin width data on post-menopausal women with hormone receptor-positive DCIS.9 Patients were considered eligible if lumpectomy margins were tumor-free by their enrolling institutions. All women uniformly underwent lumpectomy plus whole breast radiotherapy (LRT) and were randomly assigned to receive either adjuvant tamoxifen or anastrozole therapy for five years. Upon median follow-up of 9 years, the breast cancer-free interval favored the anastrozole arm vs the tamoxifen arm (hazard ratio [HR] 0.73 95% CI 0.56–0.96), with most of the benefit realized in women <60 years. As a secondary endpoint, IBTR rates were similar, with 55 (of 1,538) events in the tamoxifen arm and 46 (of 1,539) in the anastrozole arm (P=0.34). These latter findings presented a unique opportunity to examine the rates of IBTR as a function of margin width in this study population.

Methods

Trial Design

NSABP B-35 was a phase 3, double-blind, randomized trial comparing anastrozole to tamoxifen in post-menopausal women with estrogen and/or progesterone receptor-positive DCIS with or without lobular carcinoma in situ. This study was conducted in the USA and Canada (Suppl App II) and was registered with ClinicalTrials.gov: https://clinicaltrials.gov/study/NCT00053898. The protocol was approved by the institutional review board (IRB) at each participating institution. NSABP B-35 was conducted in accordance with the Helsinki Declaration. Written informed consent was required. Inclusion and exclusion criteria have been described previously.9 All patients underwent lumpectomy, followed by WBI at a dose of 5000-5040 cGy. Optional 1000 cGy boost was allowed. Patients were randomly assigned to either 20 mg of tamoxifen with an identical placebo of anastrozole or 1 mg anastrozole with an identical placebo of tamoxifen. Partial-breast irradiation (PBI) for those patients concurrently participating in NSABP B-39 was also allowed. Random assignment, stratified by age (<60 versus ≥60) was carried out via a web-based minimization algorithm through the central statistical center in Pittsburgh, PA. Patients were evaluated every 6 months for the first 5 years and yearly thereafter. Enrollment in B-35 began on January 6, 2003, ending June 15, 2006, with a total of 3,104 women entered: 1,552 randomly assigned to adjuvant tamoxifen and 1,552 to anastrozole.

Pathology Criteria

The B-35 study stipulated microscopic clear margins after primary lumpectomy and/or re-excision lumpectomy. No real-time central pathology review of margin status was performed. Institutional pathologists were instructed to follow the College of American Pathologists (CAP) guidelines for reporting DCIS, both for tumor size and margin status. Two H&E slides of the closest margin and a pathology review form had to be submitted within 3 months after randomization. Detailed margin width data were entered onto the protocol Pathology Form within three months after randomization by the institutional pathologist using the following definitions: positive (tumor on ink), close (<1 mm but tumor cells not transected), negative (≥1 mm), or unable to evaluate margin status. Additionally, for those cancers designated as having a negative margin, the distance of the closest surgical margin was indicated. Recurrences were centrally reviewed, and institutions were required to submit a tissue block of invasive or DCIS recurrence to the NSABP Biostatistical Center.

Objectives and Endpoints

Breast cancer-free interval was the primary endpoint of the NSABP B-35 trial.9 Any local, regional, or distant recurrences, as well as contralateral breast cancers, were defined as breast cancer failure events for this endpoint, with deaths censored.10 IBTR, both invasive and DCIS, occurring as a first event was a secondary trial endpoint.

The main objective of this analysis was to explore how surgical margin width (mm) relate to the incidence of IBTR and breast cancer events. Margin distance was dichotomized into two groups: either <1 mm vs ≥1 mm per protocol-defined margin status, or <2 mm vs ≥2 mm per consensus guidelines8. Additional objectives were to explore the relationship between margin width and other patient and tumor characteristics, and to investigate any modifying effect of these factors on the prognostic effect of margin width.

Analysis Cohort for This Study

The current analysis is based on information collected through February 28, 2015, which was used for the primary publication9 and was restricted to 3,077 (99.1%) patients for whom breast cancer follow-up data were available. Although negative margins were required for eligibility, there were 84 patients with forms completed by institutional pathologists listing their margins as positive. As margin status for these could not be determined, they were excluded from the current analysis. Also excluded from this analysis are 286 patients for whom margin status was listed as “unable to evaluate margins” or the pathology form was missing. Thus, 2,707 patients for whom margin status was verified as negative (≥1 mm) or close (<1 mm but tumor cells not transected) upon pathology review constitute our study population. Furthermore, narrowest margin width distance was specified for those whose margin status was considered negative or ≥1 mm. Among the latter subgroup, there were 161 cases where this additional measure of surgical margin distance was missing, thereby leaving a cohort of 2,546 patients for the 2mm threshold analysis (Figure 1).

Figure 1.

Figure 1.

Open in a new tab

Flow Diagram: NRG Oncology/NSABP B-35

Statistical Considerations

Frequency tables were constructed comparing patient demographic, clinical, and pathological variables by margin status. Differences by margin status groups were analyzed using either χ2 or Fisher’s exact tests. The 10-year cumulative probabilities of IBTR and of all breast cancer events accounting for competing risks were computed by margin status using the unadjusted cumulative incidence estimator.11,12 The Cox proportional hazards model for cause-specific hazards was used to estimate the hazard ratio (HR) for IBTR and all breast cancer events by margin status, as a single predictor, and additionally in multivariable models using other factors potentially related to breast cancer events.13,14 Analyses were performed among all B-35 patients with follow-up and margin information, using two margin classification cohorts: A) the 2,707 patients with known margin status (with patients grouped into <1 mm and ≥1 mm); and B) the 2,546 patients who had additional margin width information (using the width measurement to form groups of <2 mm and ≥2 mm reflecting the Consensus recommendation), as identified above and in Figure 1. As margin status was measured only in whole mm units with relatively few having margins exceeding 4 mm, analyses examining margin status on a continuous scale or seeking an alternate cut-point are not included.

Results

Details regarding the number of patients by the available margin width categories are presented in Table S1. Among the 2,205 patients who had negative margins (defined as ≥1 mm), 2,044 had detailed margin width information available. The median width among these patients was 4 mm, with 1st and 3rd quartiles of 2 mm and 9 mm, respectively. There were 1,667 patients with a margin width ≥2 mm and 879 patients with a margin width <2 mm, including the 502 patients who were reported as close or indefinite (<1 mm). Distributions of patient and tumor characteristics according to margin width groups (<1 mm versus ≥1 mm and <2 mm versus ≥2 mm) are shown in Table 1. Overall, characteristics of the patient subset included in this analysis are comparable to the entire trial population previously reported (data not shown).9 Importantly, no patient or disease characteristic shows strong association with margin width except for pathological tumor size. Smaller tumors were associated with larger margin widths. Of note, tumor size was estimated for 34% of participants by the local pathologist, because the tumor was not directly measurable either grossly or microscopically, and tumor size (directly measured or estimated) was unknown for 3% of patients. Treatment compliance did not differ between margin width groups. Overall, 65% of participants included in this analysis completed five years of endocrine therapy, and the median duration of treatment was 59.8 months.

Table 1.

Patient and Tumor Characteristics by Margin Width: NRG Oncology/NSABP B-35

Patient or Tumor Characteristic <1 mm ≥1 mm Total <2 mm ≥2 mm Total
N % N % N % P-value N % N % N % P-value
Treatment Group
 Tamoxifen 271 54.0 1085 49.2 1356 50.1 0.053 455 51.8 826 49.6 1281 50.3 0.29
 Anastrozole 231 46.0 1120 50.8 1351 49.9 424 48.2 841 50.4 1265 49.7
Age (Years)
 <60 227 45.2 1039 47.1 1266 46.8 0.44 394 44.8 786 47.2 1180 46.3 0.26
 ≥60 275 54.8 1166 52.9 1441 53.2 485 55.2 881 52.8 1366 53.7
Race
 White 437 87.1 1942 88.1 2379 87.9 0.34a 765 87.0 1474 88.4 2239 87.9 0.33a
 Black 47 9.4 167 7.6 214 7.9 80 9.1 120 7.2 200 7.9
 Pacific Islander 3 0.6 9 0.4 12 0.4 5 0.6 7 0.4 12 0.5
 Asian 12 2.4 52 2.4 64 2.4 20 2.3 41 2.5 61 2.4
 Native American/Alaskan 0 0 8 0.4 8 0.3 2 0.2 6 0.4 8 0.3
 Multi-racial 2 0.4 7 0.3 9 0.3 4 0.5 4 0.2 8 0.3
 Unknown 1 0.2 20 0.9 21 0.8 3 0.3 15 0.9 18 0.7
Ethnicity
 Non-Hispanic 452 90.0 2018 91.5 2470 91.2 0.56 797 90.7 1527 91.6 2324 91.3 0.49
 Hispanic or Latino 16 3.2 62 2.8 78 2.9 24 2.7 49 2.9 73 2.9
 Unknown 34 6.8 125 5.7 159 5.9 58 6.6 91 5.5 149 5.9
Tumor evident on mammogram
 Yes 487 97.0 2137 96.9 2624 96.9 0.99b 854 97.2 1614 96.8 2468 96.9 0.76b
 No 15 3.0 66 3.0 81 3.0 25 2.8 51 3.1 76 3.0
 Unknown 0 0 2 0.1 2 0.1 0 0 2 0.1 2 0.1
Comedo necrosis
 Absent 239 47.6 1028 46.6 1267 46.8 0.79b 427 48.6 770 46.2 1197 47.0 0.20b
 Present 209 41.6 925 42.0 1134 41.9 354 40.3 715 42.9 1069 42.0
 Unknown 54 10.8 252 11.4 306 11.3 98 11.1 182 10.9 280 11.0
Tumor palpable
 Yes 43 8.6 173 7.8 216 8.0 0.59b 76 8.6 124 7.4 200 7.9 0.29b
 No 459 91.4 2030 92.1 2489 91.9 803 91.4 1541 92.4 2344 92.1
 Unknown 0 0 2 0.1 2 0.1 0 0 2 0.1 2 0.1
Pathological tumor size (cm)
 <1.0 210 41.8 1204 54.6 1414 52.2 <0.001b 401 45.6 915 54.9 1316 51.7 <0.001b
 ≥1.0 274 54.6 945 42.9 1219 45.0 452 51.4 713 42.8 1165 45.8
 Unknown 18 3.6 56 2.5 74 2.7 26 3.0 39 2.3 65 2.6
Nuclear grade
 Grade 1 (low) 96 19.1 488 22.1 584 21.6 0.33b 184 20.9 357 21.4 541 21.2 0.54
 Grade 2 (intermediate) 218 43.4 914 41.5 1132 41.8 383 43.6 688 41.3 1071 42.1
 Grade 3 (high) 188 37.5 800 36.3 988 36.5 312 35.5 620 37.2 932 36.6
 Unknown 0 0 3 0.1 3 0.1 0 0 2 0.1 2 0.1
Operative area boost
 By brachytherapy 1 0.2 13 0.6 14 0.5 0.35a,b 5 0.6 9 0.5 14 0.5 0.37a,b
 By external beam 424 84.5 1815 82.3 2239 82.7 741 84.3 1372 82.3 2113 83.0
 Not done 64 12.7 321 14.6 385 14.2 112 12.7 246 14.8 358 14.1
 Unknown 13 2.6 56 2.5 69 2.5 21 2.4 40 2.4 61 2.4
Stopped treatment early for reasons other than disease progression or death
 Yes 139 27.7 649 29.4 788 29.1 0.44 259 29.5 482 28.9 741 29.1 0.77
 No 363 72.3 1556 70.6 1919 70.9 620 70.5 1185 71.1 1805 70.9
Total 502 100.0 2205 100.0 2707 100.0 879 100.0 1667 100.0 2546 100.0
Open in a new tab
a

Fisher’s exact text

b

Unknown values were excluded

Among the 2,707 women with margin status reported, 90 experienced an IBTR as a first event: 24 (4.8%) in those with <1 mm margins and 66 (3.0%) in those with margin widths ≥1 mm. The proportions of IBTRs that were invasive were 29% and 39%, respectively (Table 2). The 10-year unadjusted cumulative incidence of IBTR events for patients with a margin of <1 mm was 5.6% compared to 4.0% for those with margins that were ≥1 mm (P=0.04; Figure 2A). The HR was 1.38 (95% confidence interval (CI) 0.85–2.26) adjusted for age, treatment, and tumor size (Table 3, 1 mm). By comparison, among the 2,546 women with known margin width, 88 experienced an IBTR as a first event: 39 (4.4%) among cases with <2 mm margins and 49 (2.9%) among those with margin widths ≥2 mm. The proportion of invasive IBTRs was 41% and 33%, respectively. The 10-year unadjusted cumulative incidence of IBTR events for patients with a margin width of <2 mm was 5.3% compared to 3.8% among those with margins that were ≥2 mm (P=0.05; Figure 2B). The HR adjusted for age, treatment, and tumor size was 1.33 (CI 0.86–2.06) (Table 3, 2mm).

Table 2.

Number of Events by Margin Width: NRG Oncology/NSABP B-35


 Outcome		 Margin Width
<1 mm (n=502) ≥1 mm (n=2,205) Total (n=2,707) <2 mm (n=879) ≥2 mm (n=1,667) Total (n=2,546)
# (%) # (%) # (%) # (%)
IBTR as the first event 24 (4.8%) 66 (3.0%) 90 39 (4.4%) 49 (2.9%) 88
All breast cancer events 37 (7.4%) 150 (6.8%) 187 70 (8.0%) 109 (6.5%) 179
  Invasive IBTR 7 (1.4%) 26 (1.2%) 33 16 (1.8%) 16 (1.0%) 32
  Noninvasive IBTR 17 (3.4%) 40 (1.8%) 57 23 (2.6%) 33 (2.0%) 56
  Distant recurrence 0 (0.0%) 10 (0.5%) 10 2 (0.2%) 6 (0.4%) 8
  Contralateral breast cancer 13 (2.6%) 73 (3.3%) 86 28 (3.2%) 54 (3.2%) 82
  Angiosarcoma in the ipsilateral breast 0 (0.0%) 1 (0.1%) 1 1 (0.1%) 0 (0.0%) 1
Open in a new tab

Figure 2. Cumulative Incidence of IBTR and All Breast Cancer Events by Margin Width: NRG Oncology/NSABP B-35.

Figure 2.

Open in a new tab

A. Cumulative incidence of IBTR by Margin Width <1 mm v ≥1 mm: NSABP B-35

B. Cumulative incidence of All Breast Cancer Events by Margin Width <1 mm v ≥1 mm: NSABP B-35

C. Cumulative incidence of IBTR by Margin Width <2 mm v ≥2 mm: NSABP B-35

D. Cumulative Incidence of All Breast Cancer Events by Margin Width <2 mm v ≥2 mm: NSABP B-35

Table 3.

Effect of Surgical Margin Width on IBTR and Breast Cancer Events: NRG Oncology/NSABP B-35

Variable IBTR All Breast Cancer Events
HR 95% CI HR 95% CI
 
Multivariable Model for Margin Status (cutoff at 1 mm)
  Margin Width ≥1.0 mm 1.00 --- 1.00 ---
<1.0 mm 1.38 (0.85–2.26) 0.97 (0.67–1.42)
  Age <60 years 1.00 --- 1.00 ---
≥60 years 0.76 (0.49–1.16) 1.20 (0.89–1.61)
  Treatment tamoxifen 1.00 --- 1.00 ---
anastrozole 0.74 (0.48–1.14) 0.70 (0.52–0.95)
  Tumor size <1.0 cm 1.00 --- 1.00 ---
≥1.0 cm 2.09 (1.34–3.26) 1.47 (1.09–1.97)
Multivariable Model for Margin Width (cutoff at 2 mm)
  Margin Width ≥2.0 mm 1.00 --- 1.00 ---
<2.0 mm 1.33 (0.86–2.06) 1.13 (0.83–1.54)
  Age <60 years 1.00 --- 1.00 ---
≥60 years 0.74 (0.48–1.13) 1.17 (0.86–1.58)
  Treatment tamoxifen 1.00 --- 1.00 ---
anastrozole 0.74 (0.48–1.14) 0.74 (0.54–1.00)
  Tumor Size <1.0 cm 1.00 --- 1.00 ---
≥1.0 cm 2.07 (1.32–3.25) 1.42 (1.05–1.92)
Open in a new tab

Evaluating all breast cancer events (including IBTR), there were 37 (7.4%) women who experienced a breast cancer event among those with <1 mm margins and 150 (6.8%) among those with margins ≥1 mm (Table 2). The unadjusted cumulative incidence of all breast cancer events at 10 years were 9.5% and 8.5%, respectively (P=0.62; Figure 2C). When considering the 2 mm grouping, there were 70 (8.0%) women who experienced a breast cancer event among those with <2 mm margins and 109 (6.5%) among those with margins ≥2 mm (Table 2). Unadjusted cumulative incidences of all breast cancer events at 10 years were 10% and 8.2%, respectively (P=0.19; Figure 2D).

There were 2,672 (98.7%) patients who received adjuvant WBI. We repeated the above analyses excluding women who did not receive radiotherapy. IBTR results were similar for the groups defined by the 1 mm and 2 mm margin grouping. The 10-year unadjusted cumulative incidence of IBTR events for patients with a margin width of <1 mm and ≥1 mm margin grouping was 5.5% and 3.9%, respectively (P=0.05). The 10-year unadjusted cumulative incidence of IBTR events for patients with a margin width of <2 mm and ≥2 mm was 5.3% and 3.7%, respectively (P=0.04). Results were also similar for total breast cancer events (data not shown). There were 2,253 (82.2%) patients who received a boost dose as a part of their breast radiotherapy treatment. All analyses were also repeated among this subgroup with similar results (data not shown).

Discussion

The current analysis was facilitated by the trial requirements that institutional pathologists were first asked to classify margin width using 1 mm as a discriminant between an indefinite and negative margin and secondly, to provide the measurement of the closest margin for those in the negative or ≥1 mm category. The latter enabled the analysis using 2 mm as the discriminant margin width size, which is recommended in the Consensus Guidelines.8 Overall, smaller margin width was associated with a statistically nonsignificant relative increase in risk of IBTR events adjusted for age, treatment, and tumor size. However, the 10-year absolute difference in cumulative incidence of IBTR of only 1.6% for women with margins ≥1mm compared to <1 mm, and 1.5% for the Consensus recommended width of ≥2 mm vs. <2 mm suggests a clinically marginal benefit for a larger surgical margin.

The NSABP experience with breast-conserving treatments in women with DCIS dates to 1985, with the opening of the landmark B-17 trial.4 In that study, patients who were randomly assigned to LRT had a relative risk (RR) of IBTR of 1.48 for uncertain/involved vs tumor-free margins. In the subsequent protocol, NSABP B-24 (1991-1994), approximately 25% of the 1,799 women who underwent LRT had involved or unknown margin status.15 The HR based on 15-year follow-up for invasive IBTR with involved/unknown vs uninvolved margins was 2.61 for women who underwent LRT/placebo but was 1.27 if adjuvant tamoxifen was administered after LRT. Interestingly, the effect of involved margins on DCIS-IBTR was less: 1.65 with LRT/placebo and 1.32 with the addition of adjuvant tamoxifen.

Clinical practice in North America follows the 2 mm margin width guideline recommended by the SSO-ASCO-ASTRO Consensus Panel for patients with DCIS receiving lumpectomy and WBI.8 Other international organizations such as the St. Gallen International Expert Consensus Conference16 and UK NICE17 have also endorsed the 2 mm margin recommendation. Tadros and colleagues18 retrospectively reviewed 1,193 patients who received adjuvant WBI with a boost to lumpectomy cavity and 298 who did not. Approximately 75% were postmenopausal and 91.9% had ≥2 mm margins, but only 44.9% received adjuvant endocrine therapy. Interestingly, their 10-year LRR with close margins vs tumor-free margins was 4.8% and 3.3% respectively, comparable to our findings. More recently, Alsafi, et al19 examined 2 mm margin width among 1,790 Korean women with DCIS who underwent lumpectomy and breast radiotherapy. The 10-year local recurrence-free survival was 96% for patients with a ≥2 mm margin and 95.2% with a negative margin of ˂2 mm, also consistent with our findings.

NSABP B-35 was the first randomized trial of DCIS to collect data prospectively on margin width in postISP]-menopausal women undergoing lumpectomy followed by WBI and adjuvant endocrine therapy, thereby providing the opportunity to revisit the importance of margin width among those with uninvolved margins. The findings of this ancillary analysis involving 2,707 post-menopausal women, are widely applicable, as approximately two-thirds of newly diagnosed DCIS cases present in women ≥50 years. Notably, the trial-level meta-analysis upon which the Consensus Guidelines are based pertains primarily to ER-positive DCIS, given that 85% of cases with known hormone receptor status were ER-positive.8 Our results are further strengthened by the prospective collection of margin information from participating surgeons and pathologists in real time across many institutions. The lack of central review for margin status may be a limitation of the study design, but it is representative of real-world pathologic assessment from clinics across the US and Canada. Importantly, this analysis does not address margin width in premenopausal women or for those with ER-negative DCIS. Although this investigation was conducted in a randomized clinical trial, margin width status was not a randomized comparison, and there may be concern regarding bias due to the exclusion of patients for which true margin status was undeterminable. Among the 370 patients excluded (84 with margins reported as positive and 286 with margin status not determined), there were 11 IBTR events, a comparable proportion to the analyzed cohort.

Clinical practice is shifting towards shorter courses of breast radiotherapy, particularly PBI approaches. It is reassuring to consider that for the 1,031 patients in the NSABP B-39/RTOG 0413 trial with ER-positive and ER-negative DCIS randomly assigned to WBI vs PBI, the 10-year cumulative incidence of IBTR was 6.5% and 6.0%, respectively (HR 1.01, 95% CI 0.61–1.68).10 These results support the use of PBI instead of WBI in patients with tumor-free margins.

Other factors such as tumor biology and tissue radiosensitivity likely play a role in the long-term local recurrence of breast cancer after breast-conserving surgery and breast radiotherapy. Multigene profiling of tumors with the Oncotype DCIS and DCISionRT scoring have been studied and validated as tools to help identify patients at higher risk for local recurrence, independent of menopausal status, who may benefit from breast radiotherapy, but such tools do not address the issue of margin width.21–24

Presently, clinical management of DCIS continues to be governed by the 2 mm margin width recommended by the DCIS Margin Consensus Guidelines Panel,8 which was based on retrospective reports. Adherence to this recommendation has impacted re-excision lumpectomy rates. Specifically, Bhutiani et al24 reported an increase of the lumpectomy re-excision rate from 20.3% to 30.1%, following the publication of the guidelines. Similarly, the mastectomy conversion rate nearly quadrupled from 2.3 to 8.3% from the pre- to post-guideline time period.

In this analysis of the NSABP B-35 margin width data, margin width was not a significant predictor of IBTR risk after adjusting for other patient and tumor factors. In conclusion, the absolute differences in local recurrence rates using the 2 mm vs the 1 mm margin widths are small. Personalized decision-making is warranted regarding the need for repeat margin re-excision based on a narrow lumpectomy margin. We believe the findings of this ancillary analysis of the NSABP B-35 prospective randomized trial that showed comparable outcomes with tamoxifen or anastrozole adjuvant therapies supports reconsideration and potential modification of lumpectomy margin width criteria to reduce margin re-excision surgeries in ER-positive post-menopausal women planning to undergo adjuvant breast irradiation and endocrine therapy.

Supplementary Material

Supplementary Appendix I
Supplementary Appendix II
CONSORT Checklist
Redacted Protocol and Consent

Key Points.

Question:

The effect of lumpectomy margin width on local recurrence rates in post-menopausal women with hormone receptor positive DCIS receiving adjuvant whole breast irradiation and endocrine therapy.

Findings:

Statistically significant differences in local recurrence rates using margin width thresholds of 1mm and 2mm were shown. The 10-year unadjusted cumulative incidence of IBTR for patients with <1 mm margins was 5.6% vs 4.0% for ≥1 mm and 5.3% for margins <2 mm vs 3.8% for ≥2 mm.

Meaning:

The absolute differences using these two discrete margin width thresholds were small, 1.6% and 1.5%.

Acknowledgments

We thank the patients and their families as well as the staff members of NSABP/NRG Oncology/SDMC. We also thank Wendy L. Rea, BA, Editorial Associate, for assistance with manuscript editing and preparation for submission and Ana Stephens, BA, Protocol Development & Scientific Publications Associate for graphical and publication assistance, both of whom are employees of NSABP Foundation. They were not compensated beyond their normal salaries for this work.

Support:

Research reported herein was supported by the National Cancer Institute of the National Institutes of Health (NIH) under Award #s: U10-80868, −180822, UG1CA189867.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This manuscript is the result of funding in whole or in part by the National Institutes of Health (NIH). It is subject to the NIH Public Access Policy. Through acceptance of this federal funding, NIH has been given a right to make this manuscript publicly available in PubMed Central upon the Official Date of Publication, as defined by NIH.

Potential Conflicts of Interest:

Irene L. Wapnir: Dr. Wapnir has nothing to disclose.

Reena S Cecchini: Dr. Cecchini has nothing to disclose.

James J Dignam: Dr. Dignam reports grants from National Cancer Institute and personal fees from Merck membership on Data Monitoring Committees outside the submitted work.

Stewart J Anderson: Dr. Anderson reports personal fees from NSABP Foundation, Inc. during the conduct of the study.

Jiahe Li: Dr. Li has nothing to disclose.

Jennifer Marie Suga: Dr. Suga reports grants from National Cancer Institute 1UG1CA189821 NCORP grant during the conduct of the study.

Adam Brufsky: Dr. Brufsky reports personal fees from Astrazeneca, Pfizer, Novartis, Lilly, Roche, Daiichi Sankyo, Merck, Agendia, Sanofi, Puma, Myriad, Gilead, Bria-Cell, and Celcuity outside the submitted work.

Judith O Hopkins: Dr. Hopkins has nothing to disclose.

Kenneth Zeitzer: Dr. Zeitzer has nothing to disclose.

Janice A Lyons: Dr. Lyons has nothing to disclose.

Alison K Conlin: Dr. Conlin reports personal fees from Astra Zeneca, personal fees from Pfizer, and personal fees from Stemline outside the submitted work.

Henry Kuerer: Dr. Kuerer reports personal fees from NEJM, Inc Associate Editor, personal fees from McGraw-Hill Professional, Inc Royalties, personal fees from Endomag, Inc Honorarium, personal fees from Elsevier, Inc Royalties, and personal fees from UpToDate, Inc Editor outside the submitted work.

Shahzad Siddique: Dr. Siddique has nothing to disclose.

Laura A Vallow: Dr. Vallow has nothing to disclose.

Kathy Albain: Dr. Albain has nothing to disclose.

Mary E Cianfrocca: Dr. Cianfrocca has nothing to disclose.

Thomas Julian: Dr. Julian has nothing to disclose.

Jean Francois Boileau:

Dr. Boileau reports grants from Roche: Institution received funding for the conduct and accrual to the NSABP B-59 trial. I am the institutional PI during the conduct of the study; Reports personal fees from Roche Advisory Board, Lilly Advisory Board, Speaker’s Bureau, Pfizer Advisory Board, Novartis Advisory Board, Speaker’s Bureau, Astra-Zeeca Advisory Board, Steering Committee, Speaker’s Bureau, Merck Advisory Board, Speaker’s Bureau, Daiichi-Sankyo Consultant, Gilead Speaker’s Bureau, and Exact Sciences Advisory Board outside the submitted work.

Eleftherios P Mamounas: Dr. Mamounas reports personal fees from Exact Sciences, Genentech, Merck, Biotheranostics, GE Healthcare, Delphi Diagnostics, AstraZeneca, Novartis, Lilly, outside the submitted work.

Norman Wolmark: Dr. Wolmark reports grants from NCI/NIH U10-80868 – Grant to institution (NRG Oncology), grants from Breast Cancer Research Foundation (BCRF) – Grant to institution (NSABP Foundation), and Travel (Self), from Genentech, during the conduct of the study.

Funding Statement:

Research reported herein was supported by the National Cancer Institute of the National Institutes of Health (NIH) under Award #s: U10-80868, −180822, UG1CA189867. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This manuscript is the result of funding in whole or in part by the National Institutes of Health (NIH). It is subject to the NIH Public Access Policy. Through acceptance of this federal funding, NIH has been given a right to make this manuscript publicly available in PubMed Central upon the Official Date of Publication, as defined by NIH.

Role of the Funder/Sponsor:

The National Cancer Institute was involved in the design and conduct of the study and review and approval of the manuscript.

No authors have been paid by any pharmaceutical companies or other agencies to write this article.

The corresponding author and biostatistician had full access to all the data in the study and have final responsibility for the decision to submit for publication.

The authors take responsibility for the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript, and the decision to submit the manuscript for publication.

Drs. Dignam, Cecchini, Anderson, and Li had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Footnotes

Meeting Presentations/Previous, related works:

This is original research. Previous, related work is as follows:

Wapnir IL, et al: Margin Width and Local Recurrence in the NRG Oncology/NSABP B-35 DCIS Lumpectomy Trial. ASBrS (American Society of Breast Surgeons); Oral presentation. Presented May 2, 2025. Abstr 1985885.

Margolese RG, et al: Anastrozole versus tamoxifen in postmenopausal women with ductal carcinoma in situ undergoing lumpectomy plus radiotherapy (NSABP B-35): A randomised, double-blind, phase 3 clinical trial. Lancet. 2016; 387 (10021) 849-856. https://pubmed.ncbi.nlm.nih.gov/26686957/

Data availability / Data sharing statement:

Individual participant data that underlie the results reported in this article, after deidentification, will generally be available within 1 year after publication and will be accessible through the NCTN Data Archive. Data will be available to researchers who wish to analyze the data in secondary studies to enhance the public health benefit of the original work. Requirements may include (but not be limited to): A research plan, a Data Use Agreement (DUA), and legally binding signatures. https://nctn-data-archive.nci.nih.gov/about-us#:~:xt=%20To%20request%20data%252C%20the%20following%20are%20required%253A

REFERENCES

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Appendix I
NIHMS2176271-supplement-Supplementary_Appendix_I.docx (27.2KB, docx)
Supplementary Appendix II
NIHMS2176271-supplement-Supplementary_Appendix_II.xlsx (22.8KB, xlsx)
CONSORT Checklist
NIHMS2176271-supplement-CONSORT_Checklist.docx (37KB, docx)
Redacted Protocol and Consent
NIHMS2176271-supplement-Redacted_Protocol_and_Consent.pdf (593.7KB, pdf)

Data Availability Statement

Individual participant data that underlie the results reported in this article, after deidentification, will generally be available within 1 year after publication and will be accessible through the NCTN Data Archive. Data will be available to researchers who wish to analyze the data in secondary studies to enhance the public health benefit of the original work. Requirements may include (but not be limited to): A research plan, a Data Use Agreement (DUA), and legally binding signatures. https://nctn-data-archive.nci.nih.gov/about-us#:~:xt=%20To%20request%20data%252C%20the%20following%20are%20required%253A

RESOURCES