Radiotherapy is utilized as a part of breast-conserving therapy (lumpectomy followed by radiation therapy) or as an adjuvant therapy to mastectomy. This chapter focuses on recommendations for the use of adjuvant radiation therapy in the treatment of lymph node positive and high-risk lymph node negative breast cancer. Lower-risk disease will be covered in other chapters. Adjuvant radiation therapy recommendations for patients with node-positive disease are equally applicable in both mastectomy and breast-conserving therapy patients in most cases. Thus, the following recommendations for postmastectomy radiation therapy can be used interchangeably with patients treated with breast-conserving therapy. Controversial management decisions in the setting of one to three positive nodes, neoadjuvant chemotherapy, positive/close mastectomy margins, and T3N0 primary tumors will be covered in this chapter.

Early randomized prospective trials in oncology addressed the role of radiation for breast cancer in the postmastectomy setting. Oslo I, Oslo II, and the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-02 are examples of early trials in which women with nonmetastatic breast cancer were randomized to adjuvant radiation or no radiation after mastectomy.^1–3 Although the specifics of fields irradiated, dose delivered, and systemic therapy administered differed markedly in these trials, a statistically significant local control benefit was consistently associated with postmastectomy radiation therapy (PMRT). However, radiation therapy was associated with improved local control, an overall survival benefit was not identified. In fact, there was a suggestion of survival detriment in B-02. The negative impact on survival from PMRT was further substantiated by a meta-analysis by Cuzick and colleagues.² In this meta-analysis, which included PMRT trials from 1949 to 1974, there was a significant decrease in 10-year OS in patients treated with PMRT (57% vs. 54%; p < 0.05).²

Seven years following the initial publication, the difference was no longer statistically significant.³ The difference in OS was attributed to excess “cardiac deaths” in the irradiated patients.

This and other meta-analyses were widely criticized. The trials studied spanned three decades and consequently, the radiation techniques were not uniform. Many would consider them inadequate or outdated today. Additionally and importantly, there was no uniform use of systemic therapy in the trials analyzed in the meta-analyses.³ Despite the valid criticisms of these meta-analyses, PMRT was generally reserved for only the most advanced cases (e.g., Haagensen’s 5-grave characteristics)⁴ and mainly viewed as a means to improve local control, not survival. However, with modern radiation techniques, the local control benefit associated with PMRT would be strengthened into a survival benefit.

Between 1997 and 1999, three randomized prospective trials of PMRT were published, the Danish Premenopausal and Postmenopausal trials (82b and 82c, respectively) and the British Columbia trial.^5–7 The Danish trials (1982 to 1989), with approximately 1400 patients each, and the smaller Canadian trial (1978 to 1985), with 318 patients, randomized women treated with mastectomy and adjuvant systemic therapy to PMRT or no PMRT. The radiation fields and doses delivered were fairly uniform and all patients received systemic therapy. As seen in the earlier studies, these modern trials showed a significant difference in local regional recurrence (LRR) rate in favor of PMRT. However, for the first time, all three trials also showed a significant improvement in OS, also in favor of radiation. These trials, in contrast to their predecessors, benefited from modern standardized radiation therapy techniques, as well as modern chemotherapy.

There was general agreement with the results of the trials with respect to patients with four or more positive lymph nodes. However, the true benefit of PMRT in patients with one to three positive lymph nodes was called into question because the LRR rates in the control arms were unusually high. In retrospective reviews of prospective Eastern Cooperative Oncology Group (ECOG) and NSABP trials the 10-year LRR rates, in patients with one to three positive lymph nodes who were treated with mastectomy, chemotherapy, and no radiation, were 13%. This stands in contrast to the high LRR rates of 30% to 33% at 10 to 15 years in these three modern PMRT trials.^8,9 Consequently, many argued that radiation appeared to be more beneficial in the modern trials than it would have, had the LRR rate been closer to the historical data (ECOG and NSABP).

The identified plausible explanations for the elevated LRR rate in the PMRT trials’ include statistical anomaly and the use of nonstandard or outdated systemic and local therapies. These modern PMRT trials used Kaplan–Meier analyses, rather than cumulative incidence or crude recurrence rate analyses, as more commonly used in U.S. trials.^8,9 This may have artificially elevated the LRR rate. However, the most commonly accepted reasons for the LRR discrepancies center on what some critics would consider the administration of outdated systemic and local therapies.

Systemic therapy consisted of CMF in two of the trials and tamoxifen in the third. CMF is no longer the most common first-line chemotherapy regimen in breast cancer. Second, tamoxifen was administered without knowledge of the patients’ estrogen/progesterone receptor (ER/PR) status and was prescribed for only 1 year, as opposed to today’s recommended 5 (+)-year course. When one considers evidence that systemic therapy can also affect local control (e.g., B-06, B-21), the argument that substandard systemic therapy may have affected LRR rate seems wholly reasonable.

With respect to local therapy, some have argued that an inadequate axillary dissection may be the source of the high rate of LRR (30% to 33%) seen in the Danish and British Columbia trials of patients with one to three positive lymph nodes. The median numbers of lymph nodes removed were 7 and 11 in these trials, respectively. In contrast, the median numbers of lymph nodes removed in the retrospective analyses of ECOG and NSABP trials, both of which reported a lower (13%) LRR rate, were 15 and 16, respectively.^8,9 Because the extent of axillary dissection has been shown to have a local therapeutic benefit, some concluded that the radiation in the modern PMRT trials compensated for a less-than-ideal surgery.⁹ Therefore, the benefit of radiation in these modern trials may be exaggerated for patients with one to three positive lymph nodes. In summary, the administration of nonstandard systemic and local therapies in these modern PMRT trials prevents many physicians from offering PMRT to breast cancer patients with one to three positive lymph nodes.¹⁰

A large phase III trial opened in the United States, in which women with one to three positive lymph nodes were randomized to postmastectomy irradiation or no irradiation following mastectomy. The hope was that this trial would end the controversy concerning PMRT in this group of patients. Unfortunately, due to poor patient accrual, the trial was closed prematurely. Trials such as the United Kingdom’s Selective Use of Postoperative Radiotherapy after Mastectomy (SUPREMO) BIG 2-04 phase III randomized trial will further evaluate the true benefit of PMRT in patients with one to three positive lymph nodes.¹¹

While data from randomized prospective trials are the gold standard used to guide future therapy recommendations, there is some literature available now that suggests that the controversy over whether or not to offer PMRT to patients with one to three positive lymph nodes may soon be ending. One such paper is a reanalysis of the Danish trials in which they evaluate only patients who had an axillary dissection with greater than eight nodes removed. In this reanalysis, the authors again reported a statistically significant OS benefit in favor of PMRT in patients with one to three positive lymph nodes (48% vs. 57% at 15 years; p = .03).¹²

Additionally, in 2006, the Early Breast Cancer Trialists’ Cooperative Group presented the results of a meta-analysis of more than 3000 women with pathologically proven breast cancer, treated with mastectomy and axillary clearance, who had one to three positive lymph nodes and were randomized to adjuvant radiation or no radiation. The majority of the patients received systemic therapy. In this meta-analysis the authors reported an absolute reduction in breast cancer mortality and all-cause mortality with PMRT of 7.6% (log rank 2p = .002) and 5.3% (log rank 2p = .05), respectively.¹³ Perhaps most intriguing is a review of the Surveillance, Epidemiology and End Results (SEER) database by Buchholz and associates.¹⁴ The authors compared 12,693 patients with one to three positive lymph nodes treated with lumpectomy and radiation with 18,902 similar patients treated with mastectomy without radiation. This analysis, with limited radiation details and subject to the common critiques of retrospective studies, revealed a 15-year breast cancer-specific survival benefit in favor of the irradiated group (80% vs. 72%; p <.001). On Cox regression analysis, modified radical mastectomy (MRM) without XRT was associated with a mortality hazard ratio (HR) of 1.25 (p <.001).

Additionally, two recent prospective randomized controlled trials, the Canadian National Cancer Institute Clinical Trials Group (NCIC-CTG) MA.20 trial and the EORTC 22922/10925, further substantiate the benefit of adjuvant PMRT in women with one to three positive lymph nodes. Although these trials were specifically designed to address the role of comprehensive nodal coverage, data from these studies provide evidence to support the use of PMRT in patients with one to three positive nodes. The NCIC-CTG MA.20 included over 1800 patients, 85% with one to three positive lymph nodes, who were treated with breast-conserving surgery and systemic or hormonal therapy. Patients were randomized to whole-breast irradiation (WBI) or WBI plus regional node irradiation (RNI) targeting the supraclavicular, infraclavicular, internal mammary lymph nodes in the first to third interspaces, and high axillary lymph nodes (level III). At 5 years, the addition of RNI was associated with a DFS benefit (89.7% vs. 84.0%; p = .003), a distant metastases DFS benefit (92.4% vs. 87%; p = .002), and a trend toward an OS benefit (92.3% vs. 90.7%; p = .07).¹⁵ Additionally, the EORTC Radiation Oncology and Breast Cancer Groups phase III trial 22922/10925 trial included over 4000 patients treated from 1996 to 2004. These patients were pathologically node positive (55.5%) or were pathologically node negative with a central or medial tumor and were randomized to WBI with or without internal mammary and medial supraclavicular (IM-MS) nodal coverage. At a median follow-up of 10.9 years, there was a reduction in both regional lymph node recurrence (4.2% vs. 2.7%) and distant metastases (19.6% vs. 15.9%) in favor of the IM-MS arm. This translated into a trend toward an OS benefit (82.3% vs. 80.7%; p = .056), and with the adjustment for stratification factors there was a statistically significant OS benefit (p = .049).¹⁶

When one considers these reports, it is reasonable to agree with the concluding statement in an editorial by Marks and coworkers, “It is time that we dispense with the artificial partitioning of patient groups with 1–3 versus 4 or more positive lymph nodes.”¹⁷ Clearly, as stated in the above editorial, relying on simply the number of positive lymph nodes may oversimplify the process by which we predict LRR and by extension the need for PMRT. To that point, there is literature to suggest that the number of positive lymph nodes in relation to the total number of nodes resected is more predictive of LRR risk than simply the number of positive lymph nodes alone.^18–20

It is likely that other factors can aid in predicting LRR risk after mastectomy. For example, researchers have shown that the 21-gene recurrence score assay (Oncotype DX), commonly used to determine the risk of distant recurrence, may also be predictive of the risk of LRR.²¹ Other factors reported to be associated with increased LRR include tumor size, positive margins, extracapsular extensions, lymphovascular invasion (LVI), response to neoadjuvant chemotherapy (NAC), age, ER/PR status, and p53 overexpression.^{5,8,9,22–26} Some of these we will discuss below. While we await evaluation of putative biologic, genetic, and clinical factors predictive of LRR such that we can judge a patient’s need for PMRT, it is perhaps prudent to seriously consider PMRT for women with one to three positive axillary lymph nodes.

The main benefits of NAC for the treatment of breast cancer are its ability to increase a woman’s chance for breast conservation and to provide in vivo chemosensitivity information.^27,28 Although this change in treatment paradigm has facilitated surgery and helped to predict outcomes in certain subsets of patients, it has also caused some confusion with respect to the role of adjuvant radiation therapy. For instance, the estimation of LRR and subsequent decision for adjuvant radiation is based on clinical and pathologic factors identified perioperatively (e.g., Haagensen’s 5-grave characteristics or clinical TNM staging) or postoperatively (extent of tumor, number of positive lymph nodes, etc.). Importantly, the estimated risk of LRR was informed by clinicopathologic factors from heretofore undisturbed/untreated tumor (i.e., prior to any systemic therapy). With the advent of NAC, and the resulting shift in treatment sequence, the peri- and postoperative clinicopathologic information obtained is no longer from an undisturbed/untreated tumor. Consequently, the ways in which clinicopathologic factors are used to predict LRR may need to be reassessed. A typical conundrum would be the patient with positive lymph nodes prior to NAC, but at the time of mastectomy had no identifiable nodal metastases. Should this pathologic N0 patient be offered PMRT?

This section will focus on multiple topics inherent to NAC such as evaluation of the axilla following NAC, candidacy for conversion from mastectomy to BCT, and adjuvant radiation therapy recommendations in the NAC setting. An important point is that the decision to offer radiation for nodal management is applied independent of surgical treatment of the breast.

NAC and Evaluation of the Axilla

As described in previous sections, evaluation of axillary lymph node status is one of the strongest prognostic factors for breast cancer patients and is used to guide adjuvant systemic and radiation therapeutic decisions. Because sentinel lymph node biopsy (SLNB) has been shown to provide an accurate assessment of nodal status and is associated with decreased morbidity compared to axillary dissection, it has been adopted as the primary staging procedure in clinical node-negative non-NAC patients.^29–31 However, NAC alters the treatment sequence so that clinicopathologic factors are no longer obtained from an undisturbed/untreated tumor perhaps recommendations for surgical evaluation of the axilla may require modification.

The recently published Sentinel Lymph Node Surgery after Neoadjuvant Chemotherapy in Patients with Node-Positive Breast Cancer (Alliance) Trial, and the Sentinel-Lymph-Node Biopsy in Patients with Breast Cancer Before and After Neoadjuvant Chemotherapy (SENTINA) Trial were designed to provide reliable data for the feasibility and accuracy of a standardized SLNB after NAC.^32,33 The Alliance Trial studied women with clinically node-positive disease treated with NAC, followed by an evaluation of the axilla with SLNB and axillary dissection in order to determine the false-positive rate of SLNB. There were 525 patients with cN1 disease who had at least two SLNs excised and went on to have a complete ALND. Of the 310 patients who did not achieve a pathologic complete response, residual nodal disease was confined to the SLNs in 108 patients, confined to the nodes removed on ALND in 39 patients, and present in nodes from both procedures in 163 patients. Thus, 39 of the 310 patients with residual nodal disease had a false-negative SLN finding, an FNR of 12.6%. However, the FNR decreased to a predetermined acceptable level when three or more SLNs were examined (p = .007; FNR, 9.1% for ≥3 SLNs vs. 21.1% for 2 SLNs). Bivariable analyses also found that the likelihood of a false-negative SLN finding was significantly decreased when the mapping was performed with the combination of blue dye and radiolabeled colloid (p = .05; FNR, 10.8% combination vs. 20.3% single agent). Multivariable logistic modeling revealed that, once the number of SLNs examined (2 vs. ≥3) were accounted for, no other factors were significant predictors of a false-negative SLNB finding.³³

Additionally, the SENTINA trial was a prospective, multicenter cohort study with four arms (A to D). For the purposes of this topic we will limit our discussion to Arm C which is most similar to those treated in the Alliance trial. Arm C, consisted of women with clinically node-positive disease (n = 592) who received NAC, and who converted to clinically node-negative disease after chemotherapy and were treated with SLNB and axillary dissection. In this group, the FNR was 14.2% (95% CI 9.9 to 19.4; 32 of 226). Similar to the Alliance Trial, the FNR was consistently less than 10% for patients who had three or more sentinel lymph nodes removed. Also, multivariate analysis showed that a combined detection procedure (radiocolloid and blue dye) compared to radiocolloid tracer alone was associated with an improved detection rate in Arm C (p = 0.046). However, it was not associated with a statistically significant reduction in FNR (p = 0.145, 8.6% vs. 16%).

In conclusion, the Alliance and SENTINA trials demonstrate that SLNB with a sampling of three or more nodes is associated with an acceptable FNR (<10%). while a sampling of fewer nodes is associated with an unacceptable FNR (>10%). Both trials support that SLNB with a combined detection procedure (radiocolloid and blue dye) result in an improved evaluation of the axilla.³⁴ Based on the results of these prospective trials, patients, who present with clinically node-positive disease, subsequently convert to clinically node-negative disease after NAC, and undergo an SLNB, may trust the results of the SLNB if three or more sentinel nodes are sampled. If fewer nodes are sampled then the FNR is unacceptably high and an ALND should be seriously considered.

NAC and PMXRT Recommendations

The recommendations for PMRT in the setting of NAC are complicated by the persistent controversy concerning patients with one to three positive lymph nodes. This issue was previously discussed in the section “One to Three Node-Positive Breast Cancer” of this chapter. The U.S. literature for PMRT in the setting of NAC is mostly derived from two large prospective randomized trials, NSABP B-18 and B-27, and serial retrospective studies from MD Anderson Cancer Center (MDA).^{27,35,36–40} In study B-18, patients were randomized to either pre- or postoperative doxorubicin and cyclophosphamide (AC).⁴⁰ In study B-27, patients were randomized to preoperative AC, preoperative AC followed by preoperative paclitaxel or preoperative AC followed by postoperative paclitaxel.³⁶ Mamounas and colleagues examined patterns and predictors or LRR, following mastectomy without PMRT, from the B-18 and B-27 trials. Significant predictors of LRR following mastectomy included clinical tumor size, clinical nodal status, and pathologic nodal status/pathologic breast tumor response.⁴¹ Because of the complexity associated with the discussion of adjuvant treatment recommendations after NAC, it is perhaps best to consider the risk of LRR, and thus the need for PMRT, first as a function of clinical stage at presentation and second, according to pathologic nodal status at the time of definitive surgery.