BCT is generally reserved for patients with tumors smaller than 5 cm. However, more important than absolute tumor size is the relationship between tumor size and breast size. The tumor must be small enough, in relation to the size of the breast, to permit the tumor to be resected with clear margins and an acceptable cosmetic result. In patients with invasive breast cancer in which the tumor-to-breast size ratio is unfavorable, preoperative chemotherapy or endocrine therapy can be used to decrease the tumor size sufficiently to permit BCT (see Chapters 54 and 55).

The 2012 National Comprehensive Cancer Network (NCCN) guidelines to contraindications for BCT requiring RT include:

Margin width is assessed by applying ink (ideally colored to reflect the individual margins) to the surface of the specimen. Margin width is the distance between tumor cells and inked surface. A negative margin is defined by “no ink on any cancer cells.” Despite the NCCN guidelines, what constitutes an adequate margin is controversial (6). In surveys, no single margin width is endorsed by more than 50% of respondents (7, 8). Many practitioners favor a margin of 2 mm or even greater. As a result, re-excision is commonly used, even with a negative margin. Part of this is due to difficulties in margin assessment. Specimen processing is not standardized and varies between institutions. Sampling error is another difficulty. Even with fastidious assessment of margins, however, only a very small percentage of the entire margin is assessed. This is illustrated in Figure 35-1. On the left is an initial section with a margin width of 2 mm. A slightly deeper section, shown on the right, was taken for special studies and showed a margin width of less than 1 mm. A meta-analysis of 21 studies reporting LR relative to margin width showed that LR was significantly greater with a positive margin than a negative margin, but increasing margin width did not significantly decrease LR (9).

It is becoming increasingly clear that factors other than margin width are the primary determinants of local control. In more recent studies, which include the routine use of adjuvant systemic therapy, substantially reduced rates of LR have been seen. This is illustrated in the 10-year rates of LR in successive National Surgical Adjuvant Breast and Bowel Project (NSABP) trials in node-negative patients treated with BCT (2). In NSABP B-13, which randomized patients with estrogen receptor (ER) negative cancer to adjuvant chemotherapy or not, 10-year LR was 15.3% without chemotherapy and only 2.6% in patients with chemotherapy. Similarly, in NSABP B-14, which randomized patients with ER positive cancer to adjuvant tamoxifen or placebo, 10-year LR was 11.0% with placebo and only 3.6% in patients treated with tamoxifen. During these trials, the NSABP used “no ink on tumor” as the definition of a negative margin. Studies examining the impact of adding trastuzumab to adjuvant chemotherapy in women with HER2-overexpressing tumors demonstrate an additional 40% reduction in the risk of LR, with a median follow-up of 1.5 to 2.0 years (10). In a retrospective study from Memorial Sloan-Kettering Cancer Center, 3-year rates of LR decreased from 7% to 1% in patients with HER2-overexpressing tumors after the addition of trastuzumab to chemotherapy (11). Updated results from Dana-Farber Cancer Institute/Brigham and Women’s Hospital, and Massachusetts General Hospital included 1,434 patients; 91% were treated with adjuvant systemic therapy (not including trastuzumab), and with a median follow-up time of 85 months, the 5-year rate of LR was 1.6% and the overall crude rate of LR was 3.1% (12). (These rates are expected
to double at a median follow-up of 10 years.) In this study, as well as several others, the main prognostic factor for LR was biologic subtype approximated by hormonal receptors, HER2 status, and histologic grade, with Luminal A = HR+, HER2-, Gr 1-2; Luminal B = HR+, HER2-, Gr 3; Luminal-HER = HR+, HER2+; HER2 = HR-, HER2+; and triple negative = HR-, HER2-. The crude rate of LR by subtype was 1.5% for Luminal A, 4.0% for Luminal B, 1.0% for Luminal-HER, 10.9% for HER2+, and 8.8% for triple-negative cancers. Age was also in the final model, but the magnitude of the effect was much smaller, with a crude rate of LR of 6.5% for the patients in the lowest age quartile (ages 23 to 46) compared with only 0.9% for patients in the highest age quartile (ages 64 to 88). Margin status was not in the final model of prognostic factors for LR. These and other data indicate that the biological features of the tumor are most important in determining the risk of LR. Studies examining the relationship between molecular subtype, as approximated by receptor status, and LR are summarized in Table 35-1 (12, 13 and 14). Although definitions of the Luminal B subtype have varied over time, the lowest rates of LR are consistently seen in patients with Luminal A cancers and the highest rates in those with triplenegative cancers (12, 13 and 14), and this relationship persists even for T1a, b and microinvasive cancers (15) and in patients receiving neoadjuvant chemotherapy (16). The higher rates of LR in patients with triple-negative cancers have raised concern that these patients might benefit from treatment with mastectomy. However, patients with triple-negative cancers have the highest LR risk after both BCT and mastectomy, and retrospective studies do not demonstrate an improvement in local control after mastectomy compared to lumpectomy and radiation, even in this more aggressive tumor subset (17, 18 and 19). In multivariable analyses controlling for both conventional prognostic factors and biologic subtype, type of surgical procedure (mastectomy vs. BCT) is not a clinically significant predictor of LR for neither high-risk triple negative patients (18) nor the lower-risk ER positive subset (20).

There are additional lines of evidence that widely negative margins are not required (in the large majority of patients) for successful BCT when routine adjuvant systemic therapy is employed. Multiple studies have demonstrated that breast MRI in patients with a known primary identifies multifocal and multicentric cancers (located at a considerable distance from the known primary) in about 11% of patients (21) and triples the age-adjusted odds ratio of an initial mastectomy rather than BCT (5); however, the available studies do not demonstrate lower rate of LR with BCT in patients receiving MRI compared to those who have not (22, 23). Additional evidence for the concept that minimizing the subclinical tumor burden is not critical for reducing LR in the current era of multimodality treatment comes from the results of the American College of Surgeons Oncology Group (ACOSOG) Z0011 trial. In that study, women undergoing breast-conserving surgery, sentinel node biopsy, and whole breast irradiation, who had metastases in one or two sentinel lymph nodes, were randomized to axillary dissection or no further axillary treatment. All patients received adjuvant systemic therapy. In spite of the finding of additional nodal metastases in 27% of the axillary dissection group, only 0.9% of patients in the sentinel node-only group experienced a first recurrence in the axilla (24).

In considering the use of margins in BCT, it is useful to note that breast cancers are very often multifocal, with at least 40% of cases having microscopic foci greater than 2 cm from the edge of the cancer. This frequency is not influenced by tumor size, and these microscopic foci are more often ductal carcinoma in situ (DCIS) than invasive cancer. A key concept in the practice of BCT is that margin evaluation (and mammography) are used to insure that there is only limited residual cancer capable of being eradicated with conventional doses of RT, but not to insure there is no cancer remaining.

Thus, the current evidence indicates that in the context of highly effective systemic therapy, no ink on tumor is a sufficient margin for the large majority of patients. However, there are some exceptions, and these include cancers with an extensive intraductal component (EIC), patients receiving preoperative chemotherapy, and pure DCIS. EIC is defined by prominent DCIS within the tumor (˜25%) and present in adjacent tissue. The presence of an EIC predicted for LR in patients treated with BCT without margin evaluation; however, EIC ceased to be a prognostic factor for LR with the routine use of margin evaluation. The presence of an EIC also predicts for patients with prominent residual disease after a gross excision of the tumor. Similar considerations exist for patients with pure DCIS, and this is supported by clinical data (25). In both situations, obtaining clearly negative margins is prudent. Finally, since tumors typically respond to preoperative chemotherapy in a honeycomb pattern rather than a concentric pattern, obtaining clearly negative margins is also prudent in the setting of preoperative chemotherapy.

Radiation treatments play an important role in successful BCT for patients with invasive breast cancer. It has been clearly demonstrated that radiation treatment of the ipsilateral breast reduces the probability of LR after lumpectomy. More importantly, meta-analyses from the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) of all the randomized prospective trials comparing BCS or mastectomy with or without postoperative RT indicate that by eradicating persistent local disease after surgery, RT use reduced the risk of
subsequent death from breast cancer. The first analysis was published in 2005 and described a 4 to 1 ratio between avoiding LR at 5 years and improving survival at 15 years. After either BCS or mastectomy, RT proportionally reduced LR by 70%. An updated analysis was published in 2011 restricted to patients treated with BCS (26). In this updated analysis, 7 trials of BCS in low-risk patients, most treated with adjuvant hormonal therapy and randomized to breast RT or not, were added to the original 10 trials in the 2005 publication for a total of 10,801 women, 3,143 deaths, and a median woman-years at risk of 9.5. Importantly, the EBCTCG moved from assessing the effect of RT on LR to its effect on first failure (or first recurrence, either LR or distant metastasis). Although commonly employed in studies on the local treatment of breast cancer, actuarial calculation of time to LR is, strictly, not statistically valid. As shown in Figure 35-2, RT proportionally reduced the annual rate of any failure (LR or distant metastases) over the first 10 years by about half (relative risk [RR] = 0.52) and proportionally reduced the annual rate of breast cancer death by about one-sixth. The absolute benefit of RT was greater in patients with the greater risk of recurrence. In node-negative patients, the absolute benefit was strongly correlated with age (inversely), tumor grade and size, and ER status, with very small absolute benefit seen in some subgroups. The updated EBCTCG analysis still demonstrates that local treatment is linked to improved long-term survival; however, the new 4:1 ratio is between the reduction in first failure at 10 years (not the reduction in LR at 5 years) and the reduction in mortality at 15 years.