Margins in breast-conserving therapy: have we lost sight of the big picture?

There are few, if any, surgical procedures for cancer treatment that are based on a foundation of high-quality evidence equivalent to that which supports the use of breast-conserving therapy (BCT). Six prospective randomized trials, now with follow-up as long as 20 years, have established without a doubt, that survival after BCT, consisting of excision of the tumor to negative margins and whole breast irradiation, is equal to survival after mastectomy. Over time, rates of local recurrence after BCT have declined steadily and are now less than 10% at 10 years of follow-up. Approximately 90% of women who undergo BCT rate their cosmetic outcome as excellent or good, and with the advent of sentinel node biopsy, BCT has become an outpatient surgical procedure that allows a rapid return to the activities of daily living. All of this is good news. However, the appropriate extent of surgical resection needed to minimize local recurrence while optimizing cosmesis is still under debate. The lack of consensus regarding what constitutes an adequate margin results in the performance of mastectomies that may not be necessary, multiple trips to the operating room for margin re-excision in a significant number of patients, and unnecessarily wide resections in others.

Morrow et al. examined surgical treatment patterns in a population-based study of 2030 women treated between August 2005 and May 2006 [1]. The mastectomy rate was 38%. In 9% of cases, women chose to undergo mastectomy, but in 13% of cases, breast-conserving surgery was attempted and deemed ‘unsuccessful’ – usually after a single lumpectomy attempt. In patients who had successful BCT, 22% required re-excision. These findings suggest that considerable variation in criteria for re-excision exist, an observation confirmed by surveys of both radiation oncologists and surgeons. In one such survey, members of radiation oncology societies in North America and Europe were asked to define a ‘negative margin’. In North America, 46% of radiation oncologists used a definition of tumor cells not touching an inked surface, 29% of tumor cells within 2 mm of ink, and 15% tumor at a distance of 5 mm or more [2]. The corresponding figures for European radiation oncologists were 28, 9 and 45%, respectively. Similar variation in the definition of what constitutes a negative margin was seen in a population-based survey of 188 surgeons in the USA who were asked what negative margin width precludes the need for re-excision of a T2 invasive cancer. Only 13% endorsed tumor not touching the ink; 25% cited a margin width greater than 2 mm; and 52% required a margin greater than 5 mm [Morrow M, Unpublished data].

When evaluating these responses, it is useful to consider the margin widths required in the trials that established the validity of BCT as a therapeutic option. The National Surgical Adjuvant Breast and Bowel Project (NSABP) B-06 trial defined tumor cells not touching an inked surface as a negative margin. After 25 years of follow-up, only 14% of women who received radiotherapy had experienced a breast recurrence [3], and this figure includes tumors in other quadrants of the breast that probably represent new primary cancers. The Milan 1 trial specified removal of the quadrant of the breast containing the tumor that, depending upon the location of the carcinoma within the breast, could result in a margin several centimeters in size or a microscopically positive margin. In spite of this, local failure was observed in only 8.8% of patients at 20 years [4]. In the Institute Gustave Roussy study, a 2 cm margin was required and a 7% rate of local recurrence was observed [5]. By contrast, only gross tumor excision was mandated in the US National Cancer Institute trial [6] and the European Organization for Research and Treatment of Cancer trials [7], with local failure rates of 22 and 20% observed at 20 years.

While there is a good consensus in the literature that positive margins (defined as tumor touching ink) are associated with higher rates of local recurrence than negative margins [8,9], there is no convincing evidence that increasing margin width is associated with a reduced risk of local failure. While this seems counterintuitive, a variety of clinical observations make this a plausible finding. Residual cancer is present in the breast in 32–63% of women with clinically and mammographically unicentric tumors, regardless of whether the margins of excision are positive or negative. This finding was demonstrated in multiple pathology studies that used serial subgross sectioning to perform a detailed examination of the breast tissue [10,11], and was used to argue against the treatment of breast cancer with anything less than mastectomy. Thus, a negative margin is not a guarantee that there is no cancer remaining in the breast, but implies that the extent of residual tumor is limited enough to be controlled by breast irradiation. When viewed in this light, it is not particularly surprising that the excision of additional margins measuring 1–10 mm in width does not correlate well with rates of local control.

Other mechanical factors also influence the likelihood that small changes in margin width are likely to be major determinants of local control. The evaluation of margins has come to be regarded as an exact science, with a single tumor cell within 1 mm of an inked surface classified as an inadequate margin by some, while 3 cm of tumor 2.1 mm from the ink is not a cause for concern. Ink can easily track along the crevices of an irregular, fatty piece of breast tissue, artificially decreasing margin width, and the likelihood of having a positive margin very much depends upon the technique of sampling that is employed. The most common method of margin evaluation, inking of the specimen surface and cutting sections perpendicular to the inked surface, is a sampling technique that has the potential to leave significant amounts of tissue unsampled [12]. In spite of this, margins declared negative using this technique are associated with a low rate of local failure. The alternative technique of margin evaluation involves shaving 1–3 mm-thick pieces off of the surface of a lumpectomy specimen and defines the presence of tumor in the shaved tissue as evidence of margin involvement [12,13]. With this approach, a positive margin could be due to tumor touching an inked surface, a definition universally accepted as a positive margin, or a tumor 3 mm away from ink, a resection considered adequate by many. The clinical implications of these differences in pathologic techniques were illustrated in a study from Memorial Sloan-Kettering Cancer Center where a switch in the technique of margin processing from perpendicular to shaved increased the incidence of positive margins from 16 to 49%. This led to more re-excisions in the shaved margin group. However, the likelihood of identifying residual carcinoma at re-excision did not differ between groups (27 vs 31%).

Tissue manipulation during specimen radiography and pathologic processing also alters margin measurement. Graham et al. observed that measurements of specimen volume and height in the pathology laboratory were 30 and 46% less than those taken in the operating room, with even greater volume loss observed in specimens that were compressed for specimen radiography [14]. A 50% decrease in specimen height will convert a 1 cm anterior and posterior margin, regarded as a wide margin by many, to a 0.25 cm margin, which would often be considered a close margin. In addition, the anterior and posterior margins are anatomically limited by the skin and the pectoral muscle, so a close margin in these areas (regardless of which definition of close is used) does not carry the same potential for residual tumor as it does for margins within the breast parenchyma.

It is time to broaden our thinking beyond the simple measurement of margin width and to consider other factors that predict the likelihood of residual tumor, the biology of the cancer itself, and the impact of other treatment modalities on the risk of local recurrence when making a decision regarding the need for re-excision. Factors that have been shown to influence the likelihood of finding residual carcinoma at re-excision include the extent of tumor at or in proximity to the margin, and the histologic tumor type [15,16]. The use of postexcision mammography to document complete removal of calcifications is another mechanical means of ensuring a minimal residual tumor burden.

It is well documented that the use of adjuvant endocrine therapy or chemotherapy more than halves the risk of local recurrence after BCT when compared with placebo or untreated controls [17,18], and the majority of women with breast cancer receive some form of adjuvant therapy today. There appears to be an additional interaction between tumor subtype, as identified by a gene expression profile, and the likelihood of local control. Nguyen et al. used estrogen receptor (ER), progesterone receptor (PR) and HER-2 status to approximate the molecular subtypes of breast cancer [19]. Patients with luminal A tumors (ER or PR positive, HER-2 negative) had a 5-year cumulative risk of local recurrence of 0.8% (95% confidence interval [CI]: 0.3–2.2%), compared with 1.5% (95% CI: 0.2–10%) for those with luminal B tumors (ER or PR positive, HER-2 positive), 8.4% (95% CI: 2.2–30%) for those with ER negative, HER-2 positive tumors, and 7.1% (95% CI: 3.0–16.0%) for basal patients (ER, PR, HER-2 all negative). In a multivariable analysis, the adjusted hazard ratio for local recurrence in the HER-2 group in the era prior to the use of adjuvant trastuzumab was 9.2, and was 7.1 for the basal subtype. Data from the randomized adjuvant trials of trastuzumab indicate that it is effective in reducing local recurrence in these patients, although this was not a statistical end point in these studies [20,21]. These data do not in any way imply that wider excision, or even mastectomy, are associated with better outcome in basal breast cancers. What they do clearly show is that rates of local recurrence are extremely low in patients who are treated with effective targeted therapy, as is the case in those with hormone receptor-positive breast cancers who receive endocrine therapy; thus providing little justification for wider excisions based solely on margin width. The Oxford overview analysis provides additional data that are helpful in considering the relative contributions of tumor biology and tumor burden in the breast to local control. After mastectomy, 8% of those with node-negative breast cancers and 27.6% of those with node-positive breast cancers treated without radiotherapy, experienced local recurrence. The corresponding figures after the addition of radiation were 3 and 7.5% [22]. In a series of patients treated with BCT in NSABP trials of systemic therapy, 10-year local failure rates ranged from 4.8 to 10% in node-negative patients and 3.5–6.5% in node-positive patients [23,24].

It is time to abandon the holy grail of a single margin width that is appropriate for all patients, and the slavish performance of re-excision because a meticulous pathology examination identifies a stray tumor cell close to a margin. Re-excision has many negative consequences. It increases the risk of postoperative infection, delays the onset of adjuvant therapy postoperatively, increases the likelihood of a poor cosmetic outcome, increases medical costs, and frightens patients, sometimes leading them to choose mastectomy in order to avoid the risk of a second surgery. It is highly likely that further improvements in local control, after BCT or mastectomy, will be dependent upon the availability of new, improved, targeted therapies. It is time for the breast cancer community to come to a consensus on which patient subgroups, based on tumor biology, age and potential residual tumor burden, may benefit from a wider excision, and to make it clear to patients that when it comes to BCT and margins, bigger is not necessarily better.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.