Consensus group
Definition of negative margin
Additional recommendations
The American Society of Breast Surgeons [19]
No ink on tumor
No further surgery if margins > 1 mm, consider re-excision on a case-by-case basis for negative margins < 1 mm
NCCN [20]
> 1 mm
Margins < 1 mm at the chest wall or skin do not mandate re-excision
NICE [21]
2 mm
New Zealand Guidelines Group [22]
2 mm
Several factors should impact re-excision in close margins (< 2 mm) including age, size, grade, presence of comedo necrosis, margin location (acceptable smaller margins for posterior/anterior margins), and extent of disease near the margin
ESMO [23]
2 mm
DCIS Growth Pattern
In attempting to better define the optimal negative margin for DCIS treated with BCS, one must consider the anatomic growth pattern of the lesion . Studies comparing the radiologic distribution of calcifications to the pathologic evaluation of DCIS have provided insight into the typical growth patterns of subsets of DCIS lesions and impact margin interpretation. Holland and Hendricks [24] examined 119 mastectomy specimens containing DCIS and found that only one specimen had true multicentric disease (defined as two tumor foci separated by at least 4 cm of uninvolved breast tissue), but that segments of DCIS within one quadrant could be extensive, with 46 % of lesions measuring larger than 3 cm. Faverly et al. [25] examined the three-dimensional growth of DCIS by injecting the ducts of 60 mastectomy specimens . They observed two distinct growth patterns with equal frequency: a continuous growth pattern with uninterrupted intraductal carcinoma spread through the ductal system and a discontinuous or multifocal growth pattern with skip lesions of normal ductal segments dispersed throughout the lesion. Differentiation of DCIS was associated with specific growth patterns, with 90 % of poorly differentiated DCIS lesions showing continuous growth and 70 % of well-differentiated DCIS having a multifocal growth pattern. Of all specimens examined, 82 % had skip lesions measuring between 0 and 5 mm, while only a small percentage (8 %) had skip lesions of > 10 mm. These studies suggest that DCIS is very rarely multicentric, but within a breast quadrant, there may be continuous growth or discontinuous growth with skip lesions. Theoretically, margin assessment would be more reliable in the continuous lesions associated with poorly differentiated DCIS. For discontinuous lesions, a margin may lie within a skip lesion of normal breast tissue, and a small negative margin may be associated with a substantial residual tumor burden.
Studies evaluating risk factors for residual disease following lumpectomy for DCIS support these anatomic concepts. Neuschatz et al. [26] looked at BCS margin status as a predictor of residual DCIS at the time of re-excision. Positive margins were significantly associated with the presence of residual DCIS on re-excision, and the rate of residual disease varied by extent of margin involvement, with the following percentages of patients having residual disease at re-excision: 85 % of extensively positive margins (greater than or equal to eight involved sections or greater than four low-power fields, LPFs), 68 % of moderately positive margins (five to seven involved sections or two to four LPFs), 46 % of minimally positive margins (two to four involved sections at one geographic edge of the specimen or one LPF), and 30 % of focally positive margins (one single microscopic focus in one section). The negative margin distance was also associated with the likelihood of finding residual DCIS at re-excision (41 % of margins 0–1 mm, 31 % of margins > 1–2 mm, and 0 % of margins > 2 mm, p < 0.0001). While there were a small number of patients with margins > 2 mm (n = 10), none of these patients had residual disease at the time of re-excision, supporting the idea that most skip lesions in DCIS span a small distance .
Randomized Controlled Trials of Breast-Conserving Therapy in DCIS
Between the late 1980s and 1990s, four randomized controlled trials studied the benefit of RT in addition to BCS for women with DCIS . However, these studies were not designed to evaluate the association of margin status and LR. Three of the four studies required negative margins with no specification of a negative margin distance beyond no ink on tumor [27–29], while the SweDCIS [30] trial did not require negative margins, and 11 and 9 % of enrolled patients had positive and unknown margin statuses, respectively. Additional details regarding margin status are available from the NSABP B-17 and European Organization for Research and Treatment of Cancer (EORTC) 10853 trials. The definition of a negative margin in NSABP B-17 was no ink on transected tumor; on central pathology review, 18 % of patients included in the study were found to have involved or uncertain margins [31]. Similarly, central pathology review of the EORTC 10853 trial found that only 21 % of patients had “free” margins when a negative margin was defined as a distance of > 1 mm from ink to DCIS or a negative re-excision [10]. In spite of this, the overall incidence of LR at 8 years of follow-up was only 22 % in the NSABP B-17 trial, and 23 % at 15 years of follow-up in the EORTC study, indicating that high rates of local control can be obtained with minimal negative margins .
The Early Breast Cancer Trialists’ Collaborative Group reviewed patient-level data on 3729 women from these four trials comparing treatment of DCIS with BCS either with or without RT. The addition of RT was associated with a 10-year absolute risk reduction in any LR (either invasive or in situ) of 15 % (13 % with RT versus 28 % without RT, 2 p < 0.00001). A statistically significant benefit of RT was seen in every cohort examined, including patients with unifocal or multifocal disease and those undergoing local excision or sector resection. Of 3355 women with data available on margin status , the 10-year rates of LR were higher for women with positive margins treated with or without RT (no RT: involved margins 44 %, negative margins 26 %; RT: involved margins 24 %, negative margins 12 %) [4]. Positive margins clearly increased the risk of LR, and women with positive margins who received RT had LR rates similar to women with negative margins treated with excision alone .
Impact of RT on Optimal Margin Width
The optimal negative margin width may differ based on whether or not RT is part of the treatment plan. In the setting of BCS followed by adjuvant RT, the goal of surgery is to remove the bulk of the carcinoma and leave, at most, a subclinical volume of microscopic disease within the breast that is likely to be controlled by RT. For women treated with excision alone, the goal of surgery is to remove all of the DCIS from the breast as most LR after treatment of DCIS is thought to arise from residual disease [32]. A prospective randomized trial assessing LR in relation to margin width in DCIS has not been conducted in patients treated with or without RT, but retrospective studies have addressed this question.
Margin Width in Patients Undergoing Lumpectomy Alone
A substantial number of women with DCIS reported to the SEER database are treated with excision alone. The choice of treatment by mastectomy, BCS plus RT, or excision alone is significantly associated with patient age, histologic grade, and tumor size [33]. In a study of patients undergoing treatment for DCIS between 1996 and 2001, patients were stratified into high, moderate, and low-risk groups based on cumulative points assigned for age (> 60, 40–60, < 40 years), grade (low or intermediate without comedo necrosis, low or intermediate with comedo necrosis, high grade), and DCIS size (< 16, 16–40, > 40 mm). In this population-based sample, 17 % of high-risk patients were treated with excision alone compared to 31 % of moderate-risk patients and 44 % of low-risk patients (p < 0.001) [33]. Although treatment with excision alone is relatively common, the appropriate patient cohort for this approach and the necessary margin width are controversial.
Perhaps the most well-known study attempting to define adequate margin width in DCIS treated with and without RT was a retrospective review by Silverstein et al., examining 469 women treated between 1972 and 1998. Adjuvant RT was given to 213, and 256 were treated with surgery alone. Allocation of RT was not randomized and was routine prior to 1989. Following that time period, physician and patient preferences guided treatment decisions. Patients were retrospectively divided into margin width cohorts measuring < 1 mm (n = 112), 1 to < 10 mm (n = 224), or ≥ 10 mm (n = 133), and rates of LR were compared. Women in the no-RT group had significantly smaller tumors (9 mm vs. 13 mm, p = 0.04). At a mean follow-up of 81 months, there were three LRs in the ≥ 10 mm cohort (2 %), with no reduction in the rate of LR with the addition of RT (p = 0.92). For margin widths 1 to < 10 mm, there was a nonsignificant trend toward improved LR with the addition of RT compared to excision alone (relative risk of LR 1.49, p = 0.24), while for margins < 1 mm, there was a significant benefit with the addition of RT compared to surgery alone (RR 2.54, p = 0.01; Table 7.2) [34].
Margin width (mm) | 8-year LR with RT (n = 256) (%) | 8-year LR without RT (n = 213) (%) | p-value |
|---|---|---|---|
< 1 | 30 | 58 | 0.01 |
1–10 | 12 | 20 | 0.24 |
> 10 | 4 | 3 | 0.92 |
From this study, the authors concluded that if a 1-cm margin was obtained, RT did not provide additional benefit in reducing the risk of LR. This study was updated to include 272 cases of DCIS treated with BCS, including 212 patients treated with BCS alone and 60 patients receiving adjuvant RT. All patients had a final margin of 10 mm or greater, and the median follow-up was 53 months. Tumors treated with excision alone were significantly smaller than those treated with BCS and RT (p = 0.02), but the probability of LR at 12 years after excision alone was 14 % compared to 3 % for excision plus RT [35] in spite of the 1-cm margins. Other studies have failed to identify a 1-cm margin as obviating the need for RT in patients with DCIS. Wong et al. [36] reported a prospective study of women with low- or intermediate-grade DCIS, measuring < 2.5 cm with final surgical margins ≥ 1 cm, who were treated with excision alone, without RT or tamoxifen. The median age at diagnosis was 51 years. All patients received a post-procedure mammogram to exclude the presence of residual calcifications. From 1995 to 2002, 158 women were enrolled on the trial, but the study closed prematurely due to the number of LR events. The estimated cumulative incidence of LR for all patients at 5 years was 9.8 %, and 15.6 % by 10 years.
To further evaluate the impact of margin width on LR in patients treated with excision alone, Wehner et al. retrospectively applied the National Comprehensive Care Network (NCCN) treatment guideline recommendations to a group of low-risk DCIS patients deemed eligible for treatment with excision alone to assess rates of LR. From a single-institution database, 205 patients were identified as age 50 years or older with DCIS ≤ 2 cm, margins ≥ 2 mm, nuclear grade 1 or 2, and treatment with excision alone. Median patient age was 59 years, median DCIS size was 8 mm, and 119 patients (58 %) had a margin width of 10 mm or more. While the 6- and 12-year probabilities of LR were low at 6.6 and 7.8 %, respectively, eight of the nine observed LRs occurred in patients with a margin width of 10 mm or greater [37].
The impact of margin width was reported as a secondary outcome in the Eastern Cooperative Oncology Group (ECOG) 5194 trial, a prospective study of LR rates following excision alone for patients with DCIS > 3 mm in size. Inclusion criteria were low- to intermediate-grade DCIS measuring ≤ 2.5 cm or high-grade DCIS measuring ≤ 1 cm, with a negative margin width of at least 3 mm. In this study, complete specimen embedding and sequential sectioning for margin evaluation were performed, and a post-procedure mammogram with no residual calcifications was required. Between 1997 and 2002, 670 eligible women enrolled: 565 with low- or intermediate-grade lesions and 105 with high-grade DCIS. While the minimum margin width for inclusion was 3 mm, 49 and 53 % of the low–intermediate- and high-grade groups, respectively, had negative margin widths > 10 mm. With a median follow-up of 6.2 years, the 5- and 7-year rates of LR for the low–intermediate-grade group were 6.1 and 10.5 %, respectively. With a median follow-up of 6.7 years, the 5- and 7-year rates of LR for the high-grade group were 15.3 and 18 %, respectively [38]. Interestingly, when margins of < 10 and ≥ 10 mm were compared within the two groups, there was no significant difference in LR rates at 5 years in either the low–intermediate-grade cohort (5.6 and 6.7 %, respectively) or in the high-grade cohort (14.8 and 15.9 %, respectively) [38]. The failure of this multi-institutional, prospective study, and the study of Wong et al., to validate the importance of a 1-cm margin in maintaining local control without RT raises significant concerns about the continued use of this standard in clinical practice.
To compare these results to those obtained in a similar cohort of women treated with RT, Motwani et al. [39] retrospectively reviewed 263 patients who met ECOG Study 5194 eligibility criteria and were treated with BCS and RT between 1980 and 2009. Median patient age was slightly younger at 55 years, and median lesion size was slightly larger at 8 mm. All patients had a minimum negative margin width of 3 mm; however, additional information on margin status is not available. At a median follow-up of 6.9 years, the 5- and 7-year rates of LR for the low–intermediate-grade group were 1.5 and 4.4 %, respectively. The LR rates at 5 and 7 years for the high-grade cohort were 2.0 and 2.0 %, respectively. In this study, regardless of grade, a minimum margin width of 3 mm resulted in a low rate of LR when RT was given.
Studies of DCIS treated with excision alone have differing minimum negative margin requirements and show significant heterogeneity in LR rates (Table 7.3). Attempts to validate the importance of a 1-cm margin in DCIS patients treated without RT have been unsuccessful, and it remains unclear what the optimal margin width is for women treated with excision alone. It is likely that there is not a “one-size-fits-all margin” for this cohort as rates of LR vary based on factors such as age, tumor size, and grade, which are not controlled for in the majority of retrospective studies that have addressed this question.
Author | Minimum margin required/margin cohorts (mm) | Additional inclusion criteria | Patients with margins ≥ 10 mm | Median patient age (year) | Median tumor size (mm) | Number of years for reported LR (year) | LR rates (%) |
|---|---|---|---|---|---|---|---|
Silverstein [34] | 1 1–10 10 | – | 93/256 (36 %) | NA | 19 8 9 | 8 | 58 20 3 |
Macdonald [35] | 10 | – | 272/272 (100 %) | a | b | 12 | 14 |
Hughes [38] | 10c 10 | Low–intermediate grade, tumor size ≤ 2.5 cm | 274/565 (48 %) | 60 | 6 | 5 | 6 7 |
10c 10 | High grade, tumor size < 1 cm | 56/105 (53 %) | 59 | 5 | 7 | 15 16 | |
Wehner [37] | 2 | Tumor size ≤ 2 cm, age ≥ 50 years, non-high grade | 119/205 (58 %) | 59 | 8 | 12 | 8 |
Wong [36] | 10 | Low–intermediate grade, tumor size ≤ 2.5 cm | 143/143 (100 %) | 51 | 8d | 10 | 16 |
Lumpectomy + Radiation Therapy
While it is not clear what margin width minimizes LR in patients with DCIS treated with excision alone, it makes intuitive sense, given the anatomy of DCIS and the goal of surgical resection, that a smaller negative margin may be adequate for patients treated with BCS and RT. Solin et al. reported 15-year outcomes from a collaborative, multi-institutional database of women with DCIS treated with BCS and RT from ten institutions between 1967 and 1985 . Margins were categorized into positive, close (≤ 2 mm), or negative (> 2 mm). On univariate analysis, there was a nonsignificant trend toward increased LR with positive/close or unknown margins compared to negative margins (10-year LR rates: 10 % for negative margins, 17 % positive/close margins, 20 % unknown margins, p = 0.16). While these LR rates are higher than more recent reports, this study represents a more historic cohort of DCIS patients, being imaged with older techniques and with only 42 % of patients presenting with mammographic abnormalities alone [40]. An updated analysis of a more contemporary patient population from this multi-institutional DCIS database reported on 422 women with mammographically detected DCIS treated with definitive BCS and RT. Again, margin status was evaluated as a predictor of LR; however, it is noted that participating institutions used varying definitions of close margins (ranging from < 1 to < 3 mm), with the majority defining close margins as < 2 mm. Fifty-three percent of patients had negative margins, 11 % close margins, 9 % positive margins, and 27 % unknown. Final margin status was significantly associated with LR, with 10-year LR rates of 24 % for positive margins, 7 % for close margins, 9 % for negative margins, and 12 % for unknown margins (p = 0.03). Overall, close and negative margins had similar local failure rates, but when stratified by age groups, younger women (< 39 years of age) with close margins had higher rates of LR (67 %) than young women with negative margins (11 %). On multivariate analysis, of all margin groups, only a positive margin was independently associated with an increased risk of LR (p = 0.023) [41]. Table 7.4 summarizes studies evaluating the risk of LR by margin status following BCS for DCIS .
Reference | Margin groups (mm) | N | HR for LR | p value | Additional treatments |
|---|---|---|---|---|---|
Bolanda [43] | 1 1–9 10 | 72 129 36 | 21 2.4 1 | < 0.001 | 24 RT 60 tamoxifen 15 both RT/tamoxifen |
Pindera [46] | 1 1 uncertain | 196 838 182 | 1.50 1 1.67 | 0.02 | 435 RT |
MacDonald [45] | Positive 11–1.9 2–2.9 3–5.9 6–9.9 10 | 32 53 20 82 39 22 197 | 1 0.61 0.58 0.21 0.35 0.20 0.07 | < 0.001 | No RT or tamoxifen |
Rudloffb [47] | 2 2 | 360 1501 | 1.73 1 | 0.002 | 870 RT 370 tamoxifen |
Vargasb [48] | 2 2 | 34 198 | 3.65 1 | 0.007 | 313 RT |
Cutulib [44] | Positive/uncertain Negative | 23 70 | 1.64 1 | No RT | |
Positive/uncertain Negative | 98 361 | 1.39 1 | 0.016 | All RT | |
Waib [49] | Positive Close Negative Uncertain | 28 16 336 20 | 4.1 1.3 1 4.2 | 0.001 0.637 0.001 | No RT or tamoxifen |
Bijkerb [42] | Positive or 1 1 | 163 578 | 1.84 1 | 0.0005 | Patients randomized to RT or no RT |
Wapnir [5] | Positive/uncertain Negative | 224 676 | 2.61 (invasive) 1.65 (DCIS) 1 | 0.001 0.05 | All RTc |
A meta-analysis by Dunne et al. [12] evaluated margin status as a risk factor for LR, including 4660 women with DCIS treated with BCS and RT with a median follow-up of 85.2 months. Studies included used heterogeneous definitions of negative and close margins, and there were variations in patient age and the use of boost dose of radiotherapy. When comparing negative to positive margins, negative margins were associated with a significant reduction in LR (OR 0.36, 95 % CI 0.27–0.47, p < 0.0001). When negative margins were compared to close margins (defined by studies as anywhere from < 1 to < 5 mm) or unknown margin status, again, negative margins were significantly associated with reduced rates of LR (negative compared to close margin OR 0.59, 95 % CI 0.42–0.83, p < 0.001; negative compared to unknown margin OR 0.56, 95 % CI 0.36–0.87, p < 0.01). Close margins significantly reduced the rate of LR compared to positive margins (OR 0.43, 95 % CI 0.24–0.77, p < 0.01). However, when actual rates of LR were compared for different margin distances, small absolute differences were noted between groups. Rates of LR by margin distance were 9 % with no ink on tumor, 10 % with 1-mm margin, 6 % with a 2-mm margin, and 4 % with ≥ 5-mm margin (Table 7.5). Margins of 5 mm or greater were found to have lower rates of LR compared to margins of 1 mm or less (no ink on tumor OR 2.56, 95 % CI 1.1–7.3, p < 0.05; 1 mm OR 2.89, 95 % CI 1.3–8.1, p < 0.05), but no significant difference was seen between the 2- and 5-mm margin groups (2 mm compared to 5 mm OR 1.51, 95 % CI 0.51–5.0, p > 0.05).
