This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Recht, A. Articles by Pfister, D. G. Search for Related Content PubMed PubMed Citation Articles by Recht, A. Articles by Pfister, D. G.

Postmastectomy Radiotherapy: Clinical Practice Guidelines of the American Society of Clinical Oncology*

From the American Society of Clinical Oncology, Alexandria, VA.

Address reprint requests to American Society of Clinical Oncology, Health Services Research Department, 1900 Duke St, Ste 200, Alexandria, VA 22314; email: guidelines{at}asco.org

	ABSTRACT

TOP ABSTRACT INTRODUCTION PRACTICE GUIDELINES METHODS REFERENCES

 
OBJECTIVE: To determine indications for the use of postmastectomy radiotherapy (PMRT) for patients with invasive breast cancer with involved axil-lary lymph nodes or locally advanced disease who receive systemic therapy. These guidelines are intended for use in the care of patients outside of clinical trials.

POTENTIAL INTERVENTION: The benefits and risks of PMRT in such patients, as well as subgroups of these patients, were considered. The details of the PMRT technique were also evaluated.

OUTCOMES: The outcomes considered included freedom from local-regional recurrence, survival (disease-free and overall), and long-term toxicity.

EVIDENCE: An expert multidisciplinary panel reviewed pertinent information from the published literature through July 2000; certain investigators were contacted for more recent and, in some cases, unpublished information. A computerized search was performed of MEDLINE data; directed searches based on the bibliographies of primary articles were also performed.

VALUES: Levels of evidence and guideline grades were assigned by the Panel using standard criteria. A "recommendation" was made when level I or II evidence was available and there was consensus as to its meaning. A "suggestion" was made based on level III, IV, or V evidence and there was consensus as to its meaning. Areas of clinical importance were pointed out where guidelines could not be formulated due to insufficient evidence or lack of consensus.

RECOMMENDATIONS: The recommendations, suggestions, and expert opinions of the Panel are described in this article.

VALIDATION: Seven outside reviewers, the American Society of Clinical Oncology (ASCO) Health Services Research Committee members, and the ASCO Board of Directors reviewed this document.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION PRACTICE GUIDELINES METHODS REFERENCES

 
THE MANAGEMENT of patients with invasive breast cancer has changed substantially over the past few decades. A large proportion of such patients is now treated with breast-conserving surgery rather than mastectomy. Increasing numbers of patients (including those with histologically negative axillary lymph nodes) receive systemic therapy. However, there are still many women in North America and elsewhere who require or choose mastectomy as their primary surgical treatment.

There are several reasons or end points that might justify the use of postmastectomy radiotherapy (PMRT) for patients with invasive breast cancer. These include a reduction in the risk of local-regional failure (LRF), with its potential physical and psychologic morbidity, as well as a reduction in the risks of distant relapse and death. There is little doubt that PMRT substantially reduces the risk of LRF (usually defined as the appearance of tumor on the ipsilateral chest wall or in mastectomy scars and/or in the ipsilateral supraclavicular nodes, infraclavicular nodes, axillary nodes, the interpectoral nodes, axillary soft tissue, or internal mammary lymph nodes). However, whether PMRT directly affects the risk of distant failure and, ultimately, death due to breast cancer has been much more controversial.^1-8

This issue was highlighted by the publication in October 1997 of long-term results of the two largest trials conducted on this question for premenopausal node-positive patients treated with chemotherapy.^9,10 These studies showed that PMRT not only reduced LRF rates but also improved disease-free and overall survival rates in premenopausal patients receiving chemotherapy. However, a meta-analysis which included these as well as other trials (some of which enrolled patients treated with breast-conserving surgery or mastectomy without axillary node dissection) found that in general there was no improvement in overall survival resulting from the use of irradiation after surgery.² The applicability of the results of the Danish and British Columbia trials to patients treated in other centers and the degree of benefit obtained from irradiation in more narrowly defined patient subgroups have been debated.^8,11 Issues of concern with these two trials include the very high rates of LRF in the control arms compared with other reports (discussed in detail below) and the absence of statistically significant improvements in overall survival rates in other trials of PMRT. Also, there has been substantial disagreement regarding the optimal technical parameters of PMRT, particularly with regard to the value of specific nodal irradiation.

Therefore, in 1998, the Health Services Research Committee of the American Society of Clinical Oncology (ASCO) commissioned a panel to evaluate the evidence with regard to the value of PMRT in patients treated with systemic therapy. The Panel restricted its attention to patients with involved axillary lymph nodes or locally advanced (T4) primary tumors, as no trials have compared the role of PMRT in addition to routine systemic therapy in patients with T1 to T3 node-negative tumors.

Our report examines several aspects of this question. First, what is the evidence (and its quality and strength) regarding the impact of PMRT on LRF, disease-free survival, and overall survival? Second, is there evidence with regard to differences in such effects with regard to tumor size, the number of involved axillary nodes, and other tumor-, patient-, or treatment-related factors? Third, what information is available with regard to determining the broad technical parameters of PMRT, including its integration with systemic therapy? Finally, what are the potential side effects of PMRT, particularly in the long term, and how should these be integrated into decisions to use PMRT?

	PRACTICE GUIDELINES

TOP ABSTRACT INTRODUCTION PRACTICE GUIDELINES METHODS REFERENCES

 
"Practice guidelines are systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances."¹² Attributes of good guidelines include validity, reliability, reproducibility, clinical applicability, clinical flexibility, clarity, multidisciplinary process, review of evidence, and documentation.¹² Guidelines may be useful in producing better care and decreasing its cost. Specifically, utilization of clinical guidelines may provide (1) improvements in outcomes, (2) improvements in medical practice, (3) a means for minimizing inappropriate practice variation, (4) decision support tools for practitioners, (5) points of reference for medical orientation and education, (6) criteria for self-evaluation, (7) indicators and criteria for external quality review, (8) assistance with reimbursement and coverage decisions, and (9) criteria for use in credentialing decisions.

In formulating recommendations for radiation therapy after mastectomy, ASCO considered these tenets of guideline development, emphasizing review of data from controlled clinical trials. However, it is important to realize that guidelines cannot always account for individual variation among patients. They are not intended to supplant physician judgment with respect to particular patients or special clinical situations and cannot be considered inclusive of all proper methods of care or exclusive of other treatments reasonably directed at obtaining the same results. Accordingly, ASCO considers adherence to these guidelines to be voluntary, with the ultimate determination regarding their application to be made by the physician in light of each patient’s individual circumstances. In addition, these guidelines describe administration of therapies in clinical practice; they cannot be assumed to apply to interventions performed in the context of clinical trials, given that clinical studies are designed to test innovative and novel therapies in a disease for which better therapy is sorely needed. In that guideline development involves a review and synthesis of the latest literature, a practice guideline also serves to identify important questions for further research and those settings in which investigational therapy should be considered.

	METHODS

TOP ABSTRACT INTRODUCTION PRACTICE GUIDELINES METHODS REFERENCES

 
Panel Composition
The Panel was composed of experts in clinical medicine, clinical research, and outcomes/health services research, with a focus on expertise in breast cancer. A patient representative was also included on the Panel. The clinical experts represented all relevant medical disciplines, including surgery, medical oncology, and radiation oncology. Both academic and community practitioners were included. A steering committee under the auspices of the ASCO Health Services Research Committee chose panel participants for the clinical practice guidelines development process. Panel participants are listed in the Appendix.

Process Overview
In evaluating the evidence regarding the role of PMRT, the Panel followed the process for guidelines development established by the Canadian Medical Association.¹² The process included a systematic weighting of the level of the evidence and a systematic grading of the evidence for making a recommendation ( Table 1). ^13,14

Development Based on Evidence
The entire panel met twice. The first meeting was intended to identify topics to be addressed by the guideline, to develop a strategy for completion of the guideline, and to do a preliminary review of the initial literature search; the second meeting was intended to review the developed guideline and to evaluate more critically the recommendations and supporting evidence. The guidelines were circulated in draft form, and all members of the Panel had an opportunity to comment on the levels of evidence as well as the systematic grading of the data supporting each recommendation. Final text editing was performed by Stephen Edge and Abram Recht.

Guideline and Conflict of Interest
The content of the guidelines and the manuscript were reviewed and approved by the Health Services Research Committee and by the ASCO Board of Directors before dissemination. All members of the expert panel complied with ASCO policy on conflict of interest, which requires disclosure of any financial or other interest that might be construed as constituting an actual, potential, or apparent conflict. Members of the expert panel completed ASCO’s disclosure form and were asked to reveal ties to companies developing products that might potentially be affected by promulgation of the guidelines. Information was requested regarding employment, consultancies, stock ownership, honoraria, research funding, expert testimony, and membership on company advisory committees. The Panel made decisions on a case-by-case basis as to whether an individual’s role should be limited as a result of a conflict. There were no conflicts of interest requiring such limitations.

Revision Dates
At annual intervals, the Panel chairs and two panel members designated by the chairs will determine the need for revisions to the guidelines based on an examination of current literature. The entire panel will be reconvened every 3 years to discuss potential changes, or more frequently, if new information suggests that more timely modifications may be warranted. When appropriate, the Panel will recommend revised guidelines to the Health Services Research Committee and the ASCO Board for review and approval.

Definition of Terms
The Panel used specific language in the guidelines to reflect the type and strength of evidence available (defined according to Table 2). The term "recommendation" was used for guidelines based on level I or II evidence. The term "suggestion" was used for guidelines based on level III, IV, or V evidence, where there was panel consensus on the interpretation of the evidence. The phrase "there is insufficient evidence on which to base a guideline" was used when there was either little or no evidence on the practice in question or a lack of panel consensus on the interpretation of existing evidence or both.

SUMMARY OF THE PANEL’S FINDINGS
The randomized trials clearly show that PMRT substantially reduces the risk of LRF in patients with involved axillary nodes undergoing modified radical mastectomy who receive systemic therapy. They also show, though not as uniformly, that PMRT improves disease-free and overall survival rates. For the population of node-positive patients taken as a whole, the magnitude of these improvements is sufficient that they substantially outweigh the potential long-term risks of life-threatening complications from PMRT performed using modern radiotherapy techniques.

Nonetheless, our findings do not imply that all patients with positive axillary nodes should receive PMRT. The Panel believes the evidence shows that, as for systemic therapy, PMRT will likely cause comparable proportional reductions in LRF for various patient subgroups, and hence similar proportional improvements in disease-free and overall survival. Treatment decisions should be based on absolute differences in outcome, which may vary depending on factors such as tumor size, number of involved axillary nodes, and others.

There is much less information available as to the absolute benefits of PMRT in specific subgroups of patients. Data on changes in outcome resulting from PMRT for subgroups of patients from randomized trials are limited.

The decision as to what may constitute "sufficient" benefit to justify systemic or radiotherapeutic treatment from the breast cancer patient’s viewpoint has been examined in several studies.^20-24 Substantial differences probably exist between patients’ attitudes in this regard and that of other lay people or health-care professionals.²⁵ Therefore, we did not believe it appropriate for us to set a "threshold" value of LRF or of improvements in disease-free or overall survival rates which should trigger a decision to use PMRT.

The Panel found that the weight of the evidence from randomized trials was sufficient to recommend the routine use of PMRT for patients with four or more positive axillary lymph nodes. It is much less certain that the benefits of PMRT are sufficient to justify its use in most patients with T1/2 tumors with one to three positive nodes. The available evidence is insufficient to make recommendations for this subgroup. Further randomized trials (such as a recently opened intergroup trial in North America) are justifiable for this subgroup of patients. There are few data from randomized trials for patients with T3 or locally advanced (stage III) operable cancers, but the evidence from retrospective studies is sufficient for the Panel to suggest that PMRT be routinely used for such patients. While the consensus of the Panel was to suggest the use of PMRT for most patients treated with neoadjuvant systemic therapy, the Panel could not find sufficient evidence to determine whether all patients should be irradiated after neoadjuvant systemic therapy.

The Panel did not find sufficient evidence regarding the impact of other tumor-related, patient-related, or treatment-related factors to make recommendations or suggestions for modifying these guidelines. For example, although PMRT is commonly used for patients with close or positive margins, the data supporting this practice are fragmentary and sometimes contradictory. The 1995 and 2000 Oxford Overviews of radiotherapy^2,3 suggest that patients 50 and older may not derive a net benefit from PMRT, due to an increased risk of non–breast cancer mortality. This effect was not seen in individual trials of PMRT in which this factor was examined, however. Further investigation of this issue is needed.

The Early Breast Cancer Trialists’ Collaborative Group conducted two meta-analyses which did not show an advantage in overall survival at 10 years² and 20 years³ for patients who received radiotherapy after surgery compared with those who did not. Although in both studies there was a statistically significant reduction in the risk of breast cancer–related deaths, it was counterbalanced by an increase in the risk of non–breast cancer mortality. The findings of these Overviews and those of this panel may diverge over particular points for a number of reasons. First, the populations of concern in these two efforts were somewhat different. The Overviews examined trials in which radiation therapy was given after breast-conserving surgery, simple mastectomy, radical mastectomy, or modified radical mastectomy. The Panel dealt solely with PMRT for patients treated with mastectomy with axillary dissection. The Overviews included trials in which systemic therapy was routinely given to patients and trials in which systemic therapy was not used. The Panel restricted its review to trials of PMRT containing systemic therapy. Finally, the Overviews did not separately address the role of PMRT in women with differing number of positive nodes or other prognostic factors, whereas the Panel assessed the presence and size of an effect of PMRT in precisely such narrower subgroups.

The extent of radiation fields to be used for PMRT, when utilized, is controversial. Since the chest wall is the site at greatest risk of recurrence, we suggest that adequately treating the chest wall is mandatory. With regard to irradiation of regional lymph nodes, the situation is less clear. However, because the risk of axillary recurrence after a complete or level I/II dissection is very low, and because the combination of axillary dissection and full axillary irradiation markedly increases the risk of lymphedema, the Panel suggests that axillary radiotherapy not be given routinely to patients undergoing complete or level I/II axillary dissection. There are insufficient data to suggest subgroups that might be exceptions to such a policy. The incidence of clinical supraclavicular failure is sufficiently great in patients with four or more positive axillary nodes that we suggest supraclavicular irradiation should be given to all such patients. There are insufficient data to state whether the supraclavicular nodes should or should not be irradiated for patients with one to three positive axillary nodes. Finally, there is insufficient evidence to suggest or recommend whether internal mammary nodal irradiation should or should not be used routinely. Although such treatment was routinely given in the majority of the randomized trials (including the two Danish and the British Columbia trials, which show the greatest impact of PMRT), data on the value of internal mammary nodal irradiation are limited and contradictory. When the internal mammary nodes are deliberately treated, efforts should be made to minimize the treated volumes of heart and lung. There are insufficient data for the Panel to recommend or suggest such aspects of PMRT as total dose, fraction size, the use of bolus, and the use of scar boosts.

The optimal sequencing of chemotherapy, tamoxifen, and PMRT cannot be determined from the available evidence. The expert consensus of the Panel was that chemotherapy should be started soon after surgery, and hence the start of chemotherapy should not be delayed until after PMRT. However, in cases in which prolonged chemotherapy regimens are used, the Panel could not reach consensus on whether it was better to use a "sandwich" approach or deliver all chemotherapy before PMRT or to give concurrent chemoradiotherapy. The Panel does suggest that doxorubicin not be administered concurrently with PMRT.

There are insufficient data available to make evidence-based guidelines with regard to the integration of PMRT and reconstructive surgery. Where reconstruction can be done with a low morbidity, such that systemic therapy and PMRT will not be delayed in the large majority of cases, the consensus of the Panel is that it is reasonable to perform immediate reconstruction in patients with clinical stage I or II cancers. However, there was disagreement within the Panel regarding the use of immediate reconstruction in patients who may be candidates for PMRT, particularly those with stage IIIB tumors (ie, T4 or N2 disease) and larger T3 tumors.

The incidence of toxicities from PMRT can be difficult to assess. Many effects appear only after prolonged latency periods, and their incidence may be strongly related to the details of treatment technique as well as patient factors (such as age). The risk of carcinogenesis is so low in incidence that it should not be used in making treatment decisions. The risks of radiation pneumonitis, cardiomyopathy, brachial plexopathy, and clinically significant arm edema should also be acceptably low when state-of-the-art radiotherapy techniques are used, although follow-up in patients treated with current radiotherapy techniques may be insufficient to rule out the possibility of very late toxicities. "State-of-the-art" radiotherapy uses techniques to minimize excessive exposure of the heart and lung, pays attention to integrating the extent of nodal irradiation to the extent of axillary dissection (eg, avoiding treating a "full axillary" field in patients undergoing a complete axillary dissection), and uses daily radiation fraction sizes of 2 Gy or smaller. Data from the Oxford Overviews suggest that, when older radiotherapy techniques are used, patients may have increased risks of non–breast cancer mortality that may outweigh the benefits of PMRT for older individuals.^2,3 There are few studies of this problem in patients receiving potentially cardiotoxic drugs, such as anthracyclines or trastuzumab (Herceptin;Genentech, Inc, South San Francisco, CA), as well as radiotherapy. Considering the available evidence and its limitations, however, the Panel finds sufficient evidence to suggest that serious toxicity from PMRT in most circumstances is not sufficient to outweigh its likely benefits for the groups in whom we have recommended its use when current radiotherapy techniques are used. Similarly, there are insufficient data at present in the Panel’s expert consensus to warrant not using PMRT for some subgroups of patients, such as those with BRCA1 and BRCA2 mutations, because of a fear of increased rates of radiation carcinogenesis, compared with the rest of the population.

The Panel’s guidelines are summarized in Table 2. The following sections contain a detailed review of the relevant evidence with discussion of the Panel’s rationales for its decisions.

RANDOMIZED TRIALS COMPARING SYSTEMIC THERAPY TO SYSTEMIC THERAPY PLUS PMRT
There are at least 18 randomized studies containing over 6,300 patients which have compared systemic therapy to systemic therapy plus PMRT in (predominantly) node-positive patients treated with modified radical mastectomy ( Table 3). Important aspects of each these trials, such as eligibility criteria, population characteristics, and treatment techniques, have been summarized elsewhere.^5,7,26 The results given in Table 3 are generally taken from the latest publications of these trials. For the M.D. Anderson Cancer Center trial, results from the Oxford Overview of radiotherapy trials were used,³ as those were the only detailed ones available. Results in the British Columbia trial with 2 years of additional follow-up were reported recently in abstract form.²⁷ However, only the overall survival rates were updated for the population as a whole (43% and 52% in the control and radiotherapy arms, respectively; P = .02). Two of these trials (Danish trial 82c and one of the South Sweden trials) treated patients with tamoxifen; the rest used chemotherapy. Two trials (performed in Helsinki and by the Eastern Cooperative Oncology Group) included only patients with operable clinical stage III disease.

Only one trial reported whether PMRT reduces the ultimate risk of having uncontrollable LRF.²⁸ Fifteen (4%) of the 387 patients who received adjuvant PMRT and systemic therapy developed LRF that was "persistent" despite salvage therapy, compared with 28 such patients (7%) among the 387 who did not receive initial PMRT.

The chance that a benefit in relapse-free and overall survival rates was shown seems related to trial size. In the 10 studies with fewer than 200 assessable patients (which contained a total of 1,132 assessable patients), sometimes the control arm fared better, and sometimes the radiotherapy arm did. Eight of the nine trials that contained 200 or more patients (which together contained 5,214 assessable patients) showed trends favoring the radiotherapy arm, with absolute improvements of 2% to 17% in recurrence-free and 2% to 11% in overall survival rates in the irradiated cohort. Differences in relapse-free survival rates reached the conventional level of type I statistical significance (P < .05) in five of the eight trials that contained more than 200 patients. Only three trials (the Danish Breast Cancer Group trials 82b⁹ and 82c²⁹ and the British Columbia trial²⁷) showed statistically significant improvements in the overall survival rate for the combined-modality arm. However, the power of the trials listed in Table 3 to show that such differences as were observed were statistically significant was limited, as only two of them contained more than 1,000 patients. Because of this finding, the conclusions of this panel concerning the impact of radiotherapy on relapse-free and overall survival were therefore predominantly supported by level II evidence rather than level I evidence. Although trends toward both relapse-free and overall survival improvements were seen in the large majority of the larger trials, the difference in the numbers of studies in which these differences reached statistical significance led the Panel to find the evidence to be "generally consistent" (grade B) with regard to relapse-free survival and "inconsistent" (grade C) with regard to overall survival.

Proportional improvements in relapse-free and overall survival were substantial in the two largest trials in which patients received chemotherapy. In the Danish trial 82b, radiotherapy reduced the odds of any recurrence or death by 41%⁹; in the British Columbia trial, the reduction in the relative risk of any recurrence at 15 years was 33%.¹⁰ The relative reductions in the risk of death due to any cause in these two trials were 29% and 26%, respectively. (In the update of the British Columbia trial, this reduction was 30%.²⁷) Reductions in other trials using chemotherapy or those using tamoxifen were not as large.

It also should be pointed out that comparison of the results of these trials is hampered by several factors. One is the substantial heterogeneity between their patient populations with regard to important prognostic factors, such as the number of positive nodes and tumor size. Subgroup analysis has been performed only for a few trials. It is therefore difficult to assess the absolute size of potential disease-free and overall survival benefits in specific patient subgroups. There are also differences in the scoring of LRF between trials. Some trials scored only patients with "isolated" LRF, some scored LRF either with or without simultaneous distant relapse, and one trial scored LRF when it occurred at any time after randomization, even after the development of distant metastases.²⁸

Differing treatment factors, such as the extent of axillary dissection, may be another important source of variability in the observed results of these trials. The LRF rates in the control arms in some trials were also higher than for patients in other randomized trials as well as many retrospective studies (see below). This has led some to question the generalizability (or external validity) of their results.

There have been few meta-analyses performed of these trials. The Early Breast Cancer Trialists’ Collaborative Group conducted a meta-analysis that did not show an advantage in overall survival at 10 years² and 20 years³ for patients who received radiotherapy after surgery, compared with those who did not. However, this study included trials in which breast-conserving surgery, simple mastectomy, radical mastectomy, or modified radical mastectomy was used. Patients with both positive and negative axillary nodes were included. In addition, many of the trials used outdated radiotherapy techniques, such as the use of orthovoltage equipment. Most importantly, this study did not segregate trials in which systemic therapy was routinely given to patients from those in which systemic therapy was not used. A meta-analysis using published data from only those trials in which all patients were treated with mastectomy plus axillary dissection and also received systemic therapy showed that PMRT reduced overall mortality, with an odds ratio of 0.83 (95% confidence interval, 0.74 to 0.94; P = .004).³⁰

To conclude, PMRT reduced the risk of LRF after mastectomy in patients receiving systemic therapy by a substantial amount in all trials. The majority of available trials, particularly the larger ones, also showed that PMRT improves relapse-free and overall survival rates to a lesser, but clinically relevant, degree.

GUIDELINES FOR SPECIFIC PATIENT SUBGROUPS: Problems of Available Data and Analysis
As reviewed above, trials have shown that, for node-positive breast cancer patients as a group, aggregate, PMRT improves disease-free and overall survival in addition to local-regional control. This section will examine the evidence available to estimate the benefits from PMRT for patient subgroups defined by specific prognostic factors. As noted above, this process was hampered by limitations in the available data. Most importantly, the randomized trials of PMRT were not designed to have a high degree of statistical power to detect effects in patient subgroups, and patients were not always stratified by subgroup. None of these trials analyzed results in subgroups defined by combinations of prognostic factors, such as tumor size and the number of involved nodes. It is likely that such combinations of prognostic factors are more accurate predictors of outcome than are single factors considered in isolation.^31-35 Finally, treatment-related factors (such as details of surgery or systemic therapy) not controlled for in these trials may affect the risks of LRF substantially.

As a result of these problems, it was difficult to make conclusions with regard to the role of PMRT in specific patient subgroups. Further investigation of this area is urgently needed.

1. Patients With Four or More Positive Axillary Lymph Nodes Guideline: PMRT is recommended for patients with four or more positive axillary lymph nodes.

Level of Evidence: II

Grade of Recommendation: B

2. Patients With One to Three Positive Axillary Lymph Nodes Guideline: There is insufficient evidence to make recommendations or suggestions for the routine use of PMRT in patients with T1/2 tumors with one to three positive nodes.

Only a few of the randomized trials reviewed above have examined patient outcome based on their pathologic nodal status ( Table 4). The Danish 82b and 82c trials were the only one which included node-negative patients (who were eligible for this trial due to the presence of such "high-risk" features as invasion of the pectoralis muscle or skin by the primary tumor).^9,29 There were statistically significant differences in LRF, relapse-free survival, and overall survival rates favoring the irradiated patient arm for patients with negative, one to three positive, and four or more positive nodes in the 82b trial. In the 82c trial, there were substantial differences between the two arms in these end points also, but the statistical significance of these differences was not reported. The initial report of the British Columbia trial showed statistically significant improvements in LRF in irradiated patients in the subgroups with either one to three or four or more positive nodes.¹⁰ In an update of this trial (with unstated length of follow-up), the difference in crude LRF rates for patients with one to three positive nodes was of borderline significance between the arms (20% in the control arm and 8% in the irradiated arm, P = .066), while the difference between the arms for patients with four or more positive nodes remained highly significant (LRF rates of 51% and 17% in the two arms, respectively, P = .004).²⁷ In the initial report, a statistically significant difference in freedom from distant failure was seen only for patients with four or more positive nodes¹⁰; these rates were not updated in their recent abstract. Overall survival rates in these groups were not reported separately in their original report. In their update, there was a borderline significant improvement in patients with one to three positive nodes (53% and 64% in the control and PMRT arms, respectively, P = .07) but no significant difference in patients with four or more positive nodes (28% and 35% in the two arms, respectively, P = .2).²⁷ (There was little change in a further update of this trial restricted to patients with one to three positive nodes.³⁶) In the Scottish trial, both patients with one to three positive nodes and those with four or more positive nodes had statistically significant improvements in disease-free survival when given PMRT in addition to chemotherapy, but only patients with four or more involved nodes derived a significant advantage in cancer-specific survival from the addition of PMRT.³⁷ Three trials were restricted to patients with specific numbers of positive nodes. The Southeastern Cooperative Study Group trial showed statistically significant improvements in local-regional control and disease-free survival but no significant difference in overall survival in patients with four or more positive nodes.³⁸ Neither the trial performed for patients with one to three or (predominantly) four or more positive nodes at the Dana-Farber Cancer Institute showed differences in disease-free or overall survival.³⁹

However, there are strong reasons for questioning the generalizabilty (or external validity) of the results of these trials in these subgroups, based on retrospective (level III) data from other institutions, as well as discrepancies between the results in the randomized trials themselves. In particular, LRF rates in unirradiated patients with one to three involved nodes in the Danish 82b trial⁹ (crude rate of 30%), the Danish 82c trial²⁹ (crude rate of 31%), and the British Columbia trial¹⁰ (10-year actuarial rate of 16% but 15-year actuarial rate of 33%) were substantially higher than in the few other published series with more than 5 years of follow-up that have reported results for patients with one to three involved nodes (6% to 13%).^35,40-42 Similar discrepancies may be noted for patients with four or more positive nodes between results in these three trials (crude rates of 42% and 46% in the Danish 82b and 82c trials, and 10- and 15-year actuarial rates of and 41% and 46%, respectively, in the British Columbia trial) and those of other investigators (14% to 29%).^35,38,40-43

In summary, it is not clear that the LRF rates in the three largest trials reporting subgroup results (the Danish 82b and 82c and British Columbia trials) are representative of the results routinely achieved in North America. (Differences in surgical techniques may be responsible for this; see below.) However, in other studies, LRF rates for patients with four or more positive axillary nodes are still fairly high, even if lower than in these two trials. Although the benefits of PMRT in this subgroup may, therefore, not be quite as large as observed in the British Columbia and Danish trials, these retrospective data and the supporting data from two of the three other randomized trials lead the Panel to recommend the routine use of PMRT in patients with four or more positive axillary nodes. However, given the much lower LRF rates in retrospective studies of patients with one to three positive nodes than in the Danish and British Columbia trials, it is much less certain to the Panel whether the benefits of PMRT are sufficient to justify its routine use in most patients with one to three positive nodes. We therefore believe that the available evidence is insufficient to make recommendations or suggestions for this subgroup.

3. Patients With T3 or Stage III Tumors Guideline: PMRT is suggested for patients with T3 tumors with positive axillary nodes and patients with operable stage III tumors.

Levels of Evidence: II, III

Grade of Recommendation: C

Retrospective data suggest that the risk of LRF in unirradiated patients with operable T3 node-positive tumors is substantial (in excess of 25%), despite the use of systemic therapy and regardless of the actual number of involved axillary nodes.^35,44 Only two reports of the randomized trials of PMRT specifically analyzed results by tumor size. In the Danish 82b trial, there were statistically significant improvements in local-regional control, disease-free survival, and overall survival in patients with T1, T2, and T3 tumors.⁹ In the 82c trial, there were again substantial differences between the two arms in these end points for each tumor size, but the statistical significance of these differences was not reported.²⁹ However, the issue of whether tumor size independently affects the benefits of PMRT, controlling for nodal status, was not addressed. In several retrospective studies that did not use PMRT, the combination of tumor size plus the number of involved axillary lymph nodes was a more accurate predictor of the risk of LRF than axillary nodal status alone, but the effect of tumor size did not seem as great as that of nodal status.^31-33,35

Although the Panel decided not to address the issue of PMRT in patients with negative axillary nodes in general, we made an exception for patients with T3 primary tumors. The largest study included 101 patients treated with mastectomy without PMRT.⁴⁵ All patients had negative margins. Only 9% received systemic therapy. At a median follow-up time of 93 months, 15 patients (15%) developed LRF. In a retrospective study from Helsinki, three of five unirradiated patients had an LRF, compared with 9% (three of 33) of irradiated patients.⁴⁶ (Adjuvant systemic therapy was given to 53% of all patients in this study, but the allocation of such treatment between these two subgroups was not reported.)

Two small randomized trial of PMRT performed specifically for patients with operable locally advanced (stage III) breast cancer showed that PMRT substantially reduced the risk of LRF but did not result in statistically significant improvements in disease-free or overall survival.^47,48 The 5-year LRF rates (with or without simultaneous distant failures) among patients with such tumors in the first of these trials was 38% (five of 13) for patients receiving chemotherapy alone, compared with 7% (two of 27) for patients receiving either PMRT alone (none of 12) or chemotherapy plus PMRT (two of 15).⁴⁷ No analysis of specific patient subgroups within this stage was performed for the second trial.

Retrospective studies from individual institutions have also shown high LRF rates for operable stage III patients. In one study, the risk of LRF in patients receiving chemotherapy was 18%, compared with 8% in irradiated patients.⁴¹ In other series, the risk of LRF in patients receiving PMRT was 10% or lower.^49,50

The poor prognosis for patients with locally advanced, initially inoperable breast cancers (including inflammatory breast cancer) has led most investigators to use multimodality therapy for these patients.^51,52 Chemotherapy is used initially, and responding patients are then treated with mastectomy plus PMRT. Patients with more limited or no response often receive preoperative radiation, followed by mastectomy. There are no randomized trials addressing the role of PMRT in their treatment and only scant retrospective data on patients treated with chemotherapy and mastectomy without PMRT. In one retrospective series of patients with inflammatory breast cancer, those treated with chemotherapy plus mastectomy without radiotherapy had an LRF rate of 59% (16 of 27), compared with 15% (15 of 98) for patients treated with chemotherapy, mastectomy, and radiotherapy.⁵³ In a similar group of patients in another series, the incidence of LRF was 31% (four for 13) for unirradiated patients, compared with 7% (three of 42) for irradiated patients.⁵⁴

On balance, the Panel believed that the benefits of PMRT in improving local-regional control for patients with operable stage III cancers seem to be reasonably well established. A consistently demonstrated benefit for overall survival has not been demonstrated. Therefore, based in part on the magnitude of the risk of LRF for such patients, the Panel suggests the routine use of PMRT.

4. Patients Undergoing Preoperative Systemic Therapy Guideline: There is insufficient evidence to make recommendations or suggestions on whether all patients initially treated with preoperative systemic therapy should be given PMRT.

There are no data from randomized trials or retrospective studies on the role of PMRT in patients with clinical stage I or II tumors who undergo preoperative systemic therapy. Therefore, the use of PMRT in this setting entailed vigorous discussions among the Panel. The majority concluded that, in general, patients who require mastectomy after systemic therapy should receive PMRT. The rationale for this is based on the inability to accurately assess initial pathologic tumor size and axillary nodal status, because we recommended the use of PMRT for patients who undergo immediate surgery with four or more positive axillary nodes, and because the majority of patients who require mastectomy in this situation have more advanced cancers (clinically T3 or T4 noninflammatory cancer). However, we could not come to agreement that all patients should be treated with PMRT after neoadjuvant therapy. Further data on this issue are urgently needed, especially data on improved pretreatment staging and data correlating outcome with pathologic findings on mastectomy.

5. Modifications of these Guidelines for Special Patient Subgroups Guideline: There is insufficient evidence to make recommendations or suggestions for modifying guidelines regarding the routine use of PMRT based on other tumor-related, patient-related, or treatment-related factors.

Other Tumor-Related Characteristics
There are very few studies of how other tumor characteristics affect the risk of LRF. Such factors include the presence of vascular or lymphatic invasion,⁵⁵ tumor grade,^{9,29,33,55,56} HER2 expression,⁵⁷ P53 expression,⁵⁸ and estrogen receptor protein expression.^35,37

One potentially important consideration is the surgical margin status, or the distance of tumor from the pectoralis fascia. Although PMRT is commonly used for patients with close or positive margins, the data supporting this practice are fragmentary and sometimes contradictory.^59-61

The effect of the presence of extracapsular extension of tumor on the risk of LRF is uncertain. It did not seem to increase the risk of LRF in several series, when the number of involved nodes was accounted for.^62-65 In one study, the presence of extracapsular extension did substantially increase the risk of chest wall recurrence for patients with one to three positive axillary nodes but not for those with four or more positive nodes.⁶⁶ However, these studies were relatively small. Retrospective analysis of the British Columbia trial found that there was a substantial difference in the crude rates of overall survival for patients with one to three positive nodes when extensive extracapsular spread or extensive nodal involvement (defined as essentially replacement of the node by tumor) was found for the control arm (60% v 39% crude overall survival rates without and with this finding, respectively) but not for the PMRT arm (65% for both subgroups).²⁷ For patients with such nodal findings, the improvement in overall survival resulting from PMRT was statistically significant (P = .04). There were much smaller differences for patients with four or more positive nodes either in the control arm (overall survival rates of 41% and 25% without or with extensive extracapsular spread) or the PMRT arm (36% and 34%, respectively). However, LRF rates in these subgroups were not reported.

In summary, the Panel found insufficient evidence to recommend or suggest how other tumor characteristics should be used to modify the decisions to use or not use PMRT for patients covered in the first three guidelines. Further investigation of the pathologic and molecular correlates of LRF and treatment efficacy should be vigorously pursued.

Patient-Related Factors
Evaluation of the role of PMRT relative to patient age at diagnosis or menopausal status is hampered by the lack of routine analysis of these factors in either randomized trials or retrospective series. In the Danish 82b trial (which included only premenopausal patients), there were no differences in the risks of LRF or distant failure among patients younger than age 40 years, those 40 to 49 years old, and those age 50 or older.⁹ The first two of these age groups had similar, statistically significant improvements in 10-year disease-free and overall survival rates resulting from PMRT. In the Danish 82c trial, patients age 59 or younger and those age 60 or older also benefited similarly to each other from PMRT.²⁹ The 1995 and 2000 Oxford Overviews found that radiotherapy had almost the same proportional effects on breast cancer mortality and nonbreast deaths among women, regardless of age at diagnosis.^2,3 However, because the underlying risk of cardiac disease is higher in older patients, there was a net benefit to younger patients from radiotherapy but not for older ones. In the 1995 Overview, the odds ratios of death due to any cause (their Table 1) were 0.91 ± 0.05 for patients younger than 50 years old, 1.00 ± 0.05 for patients ages 50 to 59, and 1.01 ± 0.04 for patients age 60 and older. (Age-specific overall mortality statistics were not updated in the 2000 Overview.) Patients over 70 years of age were not generally included in either prospective or retrospective studies of chemotherapy or tamoxifen.^28,29 Hence, little can be stated about the use of PMRT in these women. Although most trials have included both pre- and postmenopausal patients, the impact of menopausal status was analyzed adequately only in the Scottish trial.³⁷

In summary, the Panel found insufficient evidence to recommend or suggest how age and menopausal status should be used to modify decisions to use or not use PMRT.

Treatment-Related Factors
Variations in surgical and systemic treatment might affect the risk of LRF and hence the value of PMRT. The Panel therefore reviewed the evidence on this subject.

Differences in surgical technique may be responsible for much of the differences in LRF rates found between different studies. In particular, the extent of axillary dissection (roughly estimated by the number of resected nodes) was more limited in a number of the randomized trials of PMRT than routinely performed by many surgeons in North America and Europe. In the Danish 82b trial, where this factor was evaluated, the extent of axillary dissection seemed to have an impact on the LRF rate in patients not undergoing PMRT.⁹ The 10-year LRF rate was 40% for 133 patients who had an axillary dissection specimen from which only zero to three nodes were recovered, compared with 32% for 511 patients in whom four to nine nodes were recovered and 27% for 211 patients in whom 10 or more nodes were examined. Of importance, LRF rates were very similar in irradiated patients, regardless of the number of recovered axillary nodes. A recent study of patients treated in four Eastern Cooperative Oncology Group (ECOG) trials of systemic therapy without PMRT showed a similar trend that was statistically significant on multivariate analysis.³⁵ The absolute impact of the number of recovered nodes on LRF rates was greater in patients with four or more positive nodes than in patients with one to three positive nodes. In the Danish 82c trial, there was no difference in the rates of LRF, relapse-free survival, or overall survival whether seven or fewer nodes or eight or more nodes were examined; however, results in patients with seven or fewer recovered nodes were not further subdivided.²⁹

The use of such findings in making decisions about the use of PMRT is complex. In the Danish trial, all three subgroups (including the one with the lowest risk of LRF) had nonsignificant trends toward improved disease-free and overall survival rates when PMRT was used.⁹ However, the size of the benefit received seemed to decrease as the risk of LRF decreased. For the subgroup of patients with 10 or more recovered nodes, the use of PMRT resulted in a 4% increase in the overall survival rate at 10 years after irradiation, compared with 10% among patients with zero to three or four to nine recovered axillary nodes. Hence, improvements in surgical technique may not eliminate the benefits of PMRT, but they may reduce it. Nonetheless, even such reduced benefits from PMRT might still be large enough to be clinically worthwhile. Few institutions have up-to-date information on LRF rates in their patients recently treated with mastectomy and systemic therapy; however, such information is needed to assess the implications of the results from the randomized trials of PMRT to their own patients.

The impact of differing systemic therapies on LRF rates has not been well studied. However, regimens that are commonly used today seem to vary little from one another. For example, cyclophosphamide, methotrexate, and fluorouracil (CMF)–based and doxorubicin-based regimens were about equally effective in this regard in the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-15 and Southwest Oncology Group 8313 trials.^67,68 In a randomized trial performed in Canada in node-positive premenopausal women, the incidence of isolated chest wall failure (as the first failure site) was approximately the same in the patients receiving standard oral CMF (8%) and patients receiving cyclophosphamide, epirubicin, and fluorouracil (CEF) (10%), whereas the risk of distant failure was significantly reduced in the CEF arm.⁶⁹ In the ECOG experience, there was little difference in LRF rates among premenopausal patients treated in trial E5181, whether doxorubicin and other agents were given in addition to CMF or not.³⁵ Whether modifications of current regimens will reduce LRF rates substantially is uncertain. Increasing the dose-intensity of cyclophosphamide did not decrease LRF rates in the NSABP B-22 trial, compared with "standard" doses.⁷⁰ Also, the risk of LRF was still substantial in unirradiated patients (three of eight), despite high-dose chemotherapy, in patients with 10 or more involved nodes in a pilot study performed at Duke University.⁷¹ There are no data yet on whether adding newer chemotherapy agents, such as the taxanes, or using altered schedules (such as "dose-dense" chemotherapy programs) will have an impact on LRF rates.

In theory, the routine addition of tamoxifen to chemotherapy for patients with estrogen receptor protein (ERP)–positive tumors might reduce the risk of LRF compared with that when chemotherapy alone is used. However, few data are available to evaluate this conjecture. The Italian Breast Cancer Adjuvant Chemo-Hormone Therapy Cooperative Group (GROCTA) trial I enrolled both pre- and postmenopausal patients with node-positive, ERP-positive tumors. The incidences of isolated LRF in patients receiving tamoxifen for 5 years, chemotherapy, or both were 12%, 13%, and 5%, respectively, at a median follow-up time of 5 years.⁷² However, these results included both patients treated with breast-conserving therapy and those treated with mastectomy; also, an unstated number of patients received PMRT.⁷³ In the ECOG experience, trials E5177 and E6177 showed trends for modest reductions in LRF rates in patients who received both chemotherapy and tamoxifen, compared with chemotherapy alone, but the differences were not statistically significant.³⁵ However, the duration of tamoxifen in these two trials (and in the two randomized trials examining the role of PMRT in patients receiving tamoxifen without chemotherapy)^28,74 was only 1 year, which would be considered inadequate today.⁷⁵ Still, longer courses of tamoxifen may not have substantial impact on the risk of LRF either. In ECOG trial 4181, ERP-positive postmenopausal women who received 1 year of tamoxifen plus chemotherapy had a 10-year risk of LRF (including patients with simultaneous distant failure) of 20%, compared with 15% for patients randomized to receive 5 years of tamoxifen in addition to the same chemotherapy. However, in ECOG trial 5181, ERP-positive premenopausal patients who received 1 year of tamoxifen plus chemotherapy had a 12% LRF rate, compared with 11% for patients who received 5 years of tamoxifen in addition to chemotherapy.³⁵ The addition of chemotherapy to tamoxifen reduced LRF rates for node-positive patients age 50 or older with ERP-positive or progesterone receptor–positive tumors in the NSABP B-16 trial.⁷⁶ However, no differences in LRF rates were seen in similar patients assigned to 2 years of tamoxifen (20%) or tamoxifen plus eight cycles of intravenous CMF (18%) in the MA.4 trial conducted by the National Cancer Institute of Canada.⁷⁷

In summary, the Panel did not find sufficient evidence to recommend or suggest how treatment recommendations for the use of PMRT should be modified according to surgical or systemic treatment factors.

Treatment Techniques: Randomized Trials Relevant to Selection of Radiotherapy Volumes
Few trials have examined the possible value of nodal irradiation in its own right, separate from that of chest wall or breast irradiation. Most of these suffer from substantial methodologic problems⁷⁸ or accrued only very small numbers of patients.^79-81 In a trial conducted in Oslo, Norway, from 1968 to 1972 (the so-called "Oslo II" trial), 265 patients were irradiated (170 patients with negative axillary nodes, 95 with positive nodes); the control group had 277 patients (186 with negative nodes, 91 with positive nodes).⁸² Local-regional recurrences occurred in 5% and 13% of the irradiated and nonirradiated groups by 10 years, respectively. These results were not subdivided by nodal status, however. There was a nonsignificant trend toward slightly worse relapse-free and overall survival rates in the patients with negative nodes who received radiotherapy, compared with the control group. The opposite trend was found in the node-positive patients (ie, favoring the radiotherapy arm), particularly for patients with medial or central tumors. (Trials that examined the value of internal mammary node dissection or irradiation are discussed below.)

Thus, there is little level I or II evidence on how selecting different treatment volumes for PMRT affects disease-free and overall survival. The Panel’s guidelines on whether such treatment should be given (and how to do it) were therefore based on level III evidence regarding the risks of pathologic involvement and of developing clinical recurrence at specific sites (recognizing the potential limitations of detecting such recurrences), weighed against the potential morbidity of such treatment (discussed in the following section). Consensus on these guidelines was generally strong, however, where the available data were consistent or generally consistent (grade A or B).

6. Chest Wall Irradiation Guideline: In patients given PMRT, we suggest that adequately treating the chest wall is mandatory.

Level of Evidence: III

Grade of Recommendation: A

7. Details of Chest Wall Irradiation Guideline: There is insufficient evidence for the Panel to recommend or suggest such aspects of chest wall irradiation as total dose, fraction size, the use of bolus, and the use of scar boosts.

The chest wall is the site at greatest risk of recurrence in patients undergoing mastectomy.⁸³ Hence, treatment of the chest wall was considered mandatory by all panel members. There are few data available to resolve questions related to more detailed technical aspects of treating the chest wall, however.

There is no agreement as to what an "adequate" or "optimal" radiotherapy regimen for PMRT is. Different centers throughout the world use very different fractionation schedules and total doses. Most institutions in the United States treat the chest wall to total doses of approximately 50 Gy in 1.8- to 2-Gy daily fractions, given five times weekly. There are no data on whether giving doses to the entire chest wall in excess of approximately 50 Gy are of additional benefit. However, the M.D. Anderson Cancer Center has used 2.5-Gy fractions to give a dose of 50 Gy in 4 weeks.⁸⁴ Institutions in Europe and Canada have often used even shorter schedules. Twice-daily treatment has also been used at the M.D. Anderson Cancer Center to treat patients with locally advanced or inflammatory carcinomas after surgery.^85,86 It is not clear whether one fractionation scheme has any advantages over another. A randomized trial of four different fractionation schemes (the START trial), which includes patients treated with PMRT, was begun in the United Kingdom in 1998. There are few data on whether or in what circumstances giving a boost dose (usually 10-16 Gy) to the mastectomy scar is of value in reducing the risk of local failure, compared with treating the entire chest wall uniformly without a boost.^87,88

The relative effectiveness of photon-based and electron-based treatment schemes was similar in the only (retrospective) study comparing the two.⁵⁴ The use of "bolus" (material with absorption of x-ray photons or electrons equivalent to that of tissue) placed on the skin surface to increase the skin dose is common, especially when photons are used to treat the chest wall. Whether it is necessary to apply bolus every day, less frequently, or at all is uncertain.^50,87,88 For patients with locally advanced disease (stage IIIB), achieving a high skin and subcutaneous dose may be more important.⁸⁵

8. Axillary Nodal Irradiation Guideline: We suggest that full axillary radiotherapy not be given routinely to patients undergoing complete or level I/II axillary dissection. There is insufficient evidence to make suggestions or recommendations as to whether some patient subgroups might benefit from axillary irradiation.

Level of Evidence: III

Grade of Recommendation: B

Full axillary irradiation was used in nearly all the randomized trials that examined the role of PMRT. However, the value of routinely giving such treatment is not clear. The risk of clinical axillary recurrence after mastectomy varies depending on whether axillary nodes are involved (and how many are positive) and by the type of axillary dissection performed.⁸⁹ In most series that used a level I/II or complete dissection of levels I to III, axillary recurrences occurred in a few percent of patients when only one to three nodes were positive, whether radiotherapy to the breast was a part of treatment or not.^35,90-92 Such failures may be more common in patients with four or more positive nodes. In one recent retrospective series, axillary recurrence occurred in 7% of patients (nine of 133) when irradiation was not used or suboptimal doses (less than 45 Gy) were given to axillary or supraclavicular nodal sites but in none of 31 patients treated with adequate irradiation doses.⁹² In a series from Lund, Sweden, no axillary failures were seen among either 46 unirradiated or 52 irradiated patients with four or more positive nodes.⁹³ In a small series of patients who underwent high-dose chemotherapy after complete dissection which revealed 10 or more positive nodes, there was no difference in axillary failure rates whether a supraclavicular field or full axillary field was treated.⁹⁴

The extent of axillary dissection, as measured roughly by the number of nodes recovered from the specimen, seems to have an impact to some degree on the risk of recurrence.⁸⁹ In a recent review of the experience of the ECOG, the 10-year cumulative incidence of axillary recurrence in patients who had one to three positive axillary nodes was 7.0% when two to five nodes were examined in the axillary dissection specimen (43 patients), 0.5% when six to 10 nodes were examined (215 patients), and 1.5% when 11 or more nodes were examined (758 patients) (P = .0009).³⁵ For patients with four or more positive nodes, the 10-year cumulative incidence of axillary recurrence was 12% when four to five nodes were examined (18 patients), 8% when six to 10 nodes were examined (138 patients), and 6% when 11 or more nodes were examined (840 patients) (P = .63).³⁵

It is not clear whether other factors play a significant role in the risk of axillary failure after surgery. The one that has been most closely examined is extracapsular extension, which does not seem to substantially increase the risk of axillary failure.^62-65,92

The Panel believed it important to note that the lower portion of the axilla (level I and part or all of level II) is ordinarily included in the same fields as are used to treat the chest wall when photon techniques are used. The so-called "supraclavicular field" (ie, lateral border at the coracoid process or medial border of the humeral head) ordinarily includes the level III nodes in most patients, as well as the true supraclavicular nodes more medially. The "full axillary field" (also called a "supraclavicular/axillary field") extends the lateral border of the supraclavicular field to split the humeral head, thus including more soft tissue laterally.⁸⁹ As discussed below, the distinction between these two fields is important with regard to the risk of lymphedema.

Therefore, despite the use of full axillary radiotherapy in the randomized trials, the Panel suggests that axillary radiotherapy not be given routinely to patients undergoing complete or level I/II axillary dissection. The available data are insufficient to suggest whether certain patient subgroups may benefit from specific axillary irradiation.

9. Supraclavicular Nodal Irradiation for Patients With Four or More Positive Axillary Lymph Nodes Guideline: The incidence of clinical supraclavicular failure is sufficiently great in patients with four or more positive axillary nodes that we suggest a supraclavicular field should be irradiated in all such patients.

Level of Evidence: III

Grade of Recommendation: A

10. Supraclavicular Nodal Irradiation for Patients With One to Three Positive Axillary Lymph Nodes Guideline: There is insufficient evidence to state whether a supraclavicular field should or should not be used for patients with one to three positive axillary nodes.

There are few surgical data on the risk of pathologic involvement of the supraclavicular nodes in patients with breast cancer. Occult supraclavicular node involvement was found in 18% (23 of 125) of patients with histologically positive axillary nodes (and none of 149 patients with negative axillary nodes) in one series in which supraclavicular node biopsies were performed routinely.⁹⁵ The risk of clinical supraclavicular recurrence after mastectomy seems to depend mainly on the extent of axillary involvement. In most series, supraclavicular recurrences occur in 1% to 4% of patients when only one to three nodes are positive.^35,90-92

Supraclavicular nodal failures are more common in unirradiated patients with four or more positive axillary nodes. In one series, supraclavicular nodal failure appeared in 17% of unirradiated or inadequately irradiated patients (17 of 102), compared with 2% of 56 irradiated patients.⁹² In another series, the risk of supraclavicular failure was 13% (six of 46) among unirradiated patients with four or more positive nodes, compared with 4% (two of 52) for those irradiated.⁹³ However, in another recent study, the incidence of supraclavicular failure was only 3% (one of 36) without irradiation.⁹⁶ Doses of 45 to 50 Gy in 1.8- to 2-Gy fractions seem adequate to achieve control in the great majority of patients.^84,89

The appropriateness of giving supraclavicular nodal irradiation was discussed extensively by the Panel. This subject has not been well studied. Some argued that such failures can be very difficult to control and can cause great morbidity.⁹⁷ Others believed that the potential morbidity of irradiation itself (discussed in detail below) might outweigh the risk of uncontrollable supraclavicular recurrence, especially in patient subgroups with relatively low failure rates. Further, there are no clear data on the impact of supraclavicular irradiation on overall survival. Two trials (one conducted by the European Organization for Research and Treatment of Cancer, the other by the National Cancer Institute of Canada) are currently open in which patients are randomized to receive or not receive both supraclavicular and internal mammary nodal irradiation. These may someday help settle this contentious issue, but at present there are clearly substantial differences of opinion on this subject within the radiation oncology community, as well as among the Panel members.

Therefore, based on these considerations of morbidity and the substantial observed incidence of clinical supraclavicular failure in patients with four or more positive axillary nodes, the Panel suggests that a supraclavicular field should be irradiated in all such patients. However, there are insufficient data for the Panel to state whether a supraclavicular field should or should not be used for patients with one to three positive axillary nodes.

11. Internal Mammary Nodal Irradiation Guideline: There is insufficient evidence to make suggestions or recommendations on whether deliberate internal mammary nodal irradiation should or should not be used in any patient subgroup.

One of the most controversial issues regarding PMRT is treatment of the internal mammary nodes. Internal mammary nodal irradiation was used in the majority of trials of PMRT, including the two Danish and the British Columbia trials. However, the data regarding the value of routine internal mammary nodal irradiation are limited and contradictory.

In older studies, the incidence of internal mammary node metastases was approximately 10% in patients with a negative axillary dissection and 20% to 50% in patients with a positive dissection.^97-100 More recent studies tend to show lower risks of involvement.^101-104 Tumor size, clinical stage, and the number of axillary nodes involved affect the risk of internal mammary node involvement.^101-105 The location of the primary tumor in the breast seems to have only a minor impact on the risk of internal mammary node involvement, when controlled for axillary nodal status and tumor size.^98,99,105

Clinical recurrence in internal mammary nodes in patients with positive axillary nodes is rare in most series, even when radiotherapy is not given,^{35,62,106,107} although a few studies have found higher rates of recurrence (9% to 10%).^108,109 In one series in which 20 patients had biopsy-proven internal mammary node involvement but did not subsequently receive irradiation, only one patient had a clinical recurrence.¹⁰⁹ The reasons for this discrepancy between the apparently large risk of pathologic involvement and this low incidence of clinical manifestation are uncertain.

One of the reasons why it may be difficult to show that specific internal mammary nodal irradiation improves outcome, at least in patients undergoing breast or chest wall irradiation, is the anatomic location of these nodes. The nodes most likely to be involved in breast cancer are in the upper three interspaces.^103,108 These tend to be located less deeply than the ones located more inferiorly. Hence, in many patients, many or most of the internal mammary nodes will likely be included in tangential photon treatment fields, even if they extend only to the midline.¹¹⁰ Thus, techniques that attempt to more comprehensively treat the internal mammary nodes may add only a small degree of benefit (if any) to standard chest wall irradiation fields.

Retrospective studies of irradiated patients differ on whether there are any benefits to giving internal mammary nodal irradiation.^111,112 A few randomized trials have focussed on whether internal mammary nodal treatment improves patient outcome. Two trials showed no improvement in survival in patients who underwent internal mammary node dissection in addition to standard radical mastectomy.^98,100,113 However, systemic therapy was not used in these trials, and hence the applicability of these results to patients receiving systemic therapy is not known. A trial performed from 1985 to 1993 at the National Cancer Institute Hospital in Tokyo, Japan, randomly allocated 150 patients with biopsy-proven internal mammary node involvement to either radical resection of the internal mammary supraclavicular chain, irradiation of the supraclavicular and internal mammary nodes, or no further surgery or deliberate irradiation of these areas.¹¹⁴ All patients were treated with quadrantectomy and irradiation of the breast, as well as six courses of CMF. The 5-year disease-free survival rates were similar in the three arms (57%, 53%, and 51%, respectively), although the risk of supraclavicular and/or internal mammary recurrence was lowest in the irradiated group (12%, 0%, and 16%, respectively). A randomized study of internal mammary irradiation performed in Tampere, Finland, between 1989 and 1991 in patients treated with breast-conserving surgery showed no difference in relapse rates at a median follow-up of 2.7 years, but only 13% of the 270 patients included had positive axillary nodes and only 18% had centrally or medially located primary tumors.¹¹⁵ A trial was conducted from 1992 to 1997 in France in which 1,391 patients were randomized to receive either chest wall and supraclavicular radiotherapy or chest wall, supraclavicular, and internal mammary irradiation after modified radical mastectomy (axillary irradiation was optional).¹¹⁶ About half of these patients had primary tumors located in the central or inner quadrants of the breast and half had tumors in any