Impact of axillary lymph node dissection on breast cancer outcome in clinically node negative patients

A Systematic Review and Meta-Analysis

Abstract

BACKGROUND:

The regional lymph node control and survival impact of axillary dissection in breast cancer has been the subject of multiple randomized trials, with various results. This study reviews and conducts a meta-analysis of contemporary trials of axillary dissection in patients with early stage breast cancer.

METHODS:

A systematic MEDLINE review identified 3 randomized trials published between January 2000 and January 2007 of axillary dissection versus no dissection in clinically lymph node negative early stage breast cancer patients. A fourth trial of axillary radiotherapy versus no axillary treatment was also identified and included in this review. Meta-analyses were performed for survival, axillary recurrence, metastatic disease, and ipsilateral breast recurrence.

RESULTS:

All trials reported a higher rate of axillary recurrence (1.5%-3%, median follow-up 5-15 years) in the absence of axillary dissection or radiotherapy. Overall survival was similar with and without definitive axillary treatment in 3 of the 4 trials, with an increased rate of nonbreast cancer-related death in the observation arm of the fourth trial. Meta-analyses found no significant difference in overall survival (odds ratio [OR] 1.55; 95% confidence interval [CI], 0.74-3.24), metastases (OR 0.91; 95% CI, 0.65-1.29), or ipsilateral breast recurrence (OR 1.11; 95% CI, 0.68-1.83) associated with axillary treatment. A significantly lower rate of axillary recurrence was seen after lymphadenectomy (OR 0.28; 95% CI, 0.11-0.73, P<.01).

CONCLUSIONS:

Axillary dissection does not confer a survival benefit in the setting of early stage clinically lymph node negative breast cancer. Although the rate of axillary failure was increased in the absence of dissection, the absolute risk was found to be extremely low. Cancer 2009. © 2009 American Cancer Society.

The impact of definitive axillary treatment on breast cancer survival in clinically lymph node negative patients has been an area of ongoing controversy. A landmark study, NSABP B-04, found no difference in overall survival or distant metastases when patients with breast cancer and clinically negative axillae were randomized to axillary dissection, axillary radiotherapy, or no axillary treatment.1 Criticisms of this study were that several axillary lymph nodes were removed in some patients without deliberate axillary dissection, and the trial was not adequately powered to detect a small difference in survival that may have been conferred by axillary treatment. Many subsequent randomized studies have sought to address this question and have found various results.

A meta-analysis published in 1999 by Orr included 6 randomized trials of axillary dissection versus no dissection and found an average survival benefit of 5.4% with the addition of axillary dissection.2 Of important note is that the majority of the trials in this meta-analysis predated the use of routine mammographic screening and other modern imaging modalities, which have now led to the detection of earlier stage cancers with a corresponding lower risk of lymph node involvement. In addition, the Orr meta-analysis included the 5-year results of the Insitut Curie trial, which showed a survival benefit for axillary dissection; this survival benefit was no longer detectable in the updated 15-year results of the trial, which will be presented in this review.

The aim of this present study was to review contemporary randomized trials of axillary treatment versus no treatment to assess whether a survival benefit for axillary lymph node treatment exists in the present day setting of predominantly early stage breast cancers, with breast conservation in the majority of patients.

MATERIALS AND METHODS

A comprehensive MEDLINE search was conducted to identify randomized trials of definitive axillary treatment (dissection or radiotherapy) published in English between January 2000 and January 2007. Search terms included “breast cancer”, “axillary treatment”, “axillary dissection”, and “survival” with the following limitations: publication date January 1, 2000, through January 1, 2007; English language; randomized controlled trial; and human subjects. A total of 92 publications were identified meeting these search criteria. Further review led to the exclusion of studies in which the primary randomization was not evaluating axillary treatment, sentinel lymph node biopsy was performed, breast conserving therapy was not allowed, or clinically lymph node positive patients were included.

One author (M.S.) performed data extraction on all trials. Collected data included total number of patients, median follow-up, type of surgery, extent of radiotherapy, median patient age, disease stage, and numbers of patients who had the following outcomes of interest: overall survival, disease-free survival, local in-breast recurrence, regional axillary lymph node recurrence, and distant metastatic failure. Where absolute numbers were not provided, best estimates were made based on the provided percentages.

Meta-analyses of odds ratios (OR) were performed using a random-effects model, which assigns a weight to each study based on both the within-study variance and the between-study heterogeneity.3 The primary endpoint for this meta-analysis was overall survival, but the additional endpoints of disease-free survival, ipsilateral local breast tumor recurrence, axillary regional lymph node recurrence, and distant metastatic recurrence were also analyzed. To improve the homogeneity of median follow-up time, the 5-year results from the Institut Curie, rather than the 15-year results, were included in the analysis.

RESULTS

This review resulted in a total of 4 trials: 2 evaluating axillary dissection versus no treatment; 1 of axillary dissection versus radiotherapy; and 1 of axillary radiotherapy versus no treatment, as detailed in Table 1.4-7

Table 1. Patient Characteristics and Study Results From the Reviewed Randomized Trials

Study	Martelli	IBCSG	Louis-Sylvestre	Veronesi
IBCSG indicates International Breast Cancer Study Group; ER, estrogen receptor; PR, progesterone receptor; NR, not reported; NS, P value not statistically significant (where missing, statistical significance was not reported).
Intervention arm	Dissection	Dissection	Dissection	Axillary radiotherapy
Control arm	No dissection	No dissection	Axillary radiotherapy	No axillary radiotherapy
Dates	1996-2000	1993-2002	1982-1987	1995-1998
Median follow-up (y)	5	6.6	15	5.3
Total patient no.	219	473	658	435
Mean/median age (y)	70	74	51	57
T1 (%)	93	56	67	100
ER+ and/or PR+ (%)	88	80	60	81.2
Tamoxifen, for any duration (%)	100	97	3.3	76.1
Chemotherapy (%)	NR	NR	4.3	6.9
Overall survival (%)	87 vs 92 NS	75 vs 73 NS	75.5 vs 73.8 NS	99 vs 95 P < .01
Disease-free survival (%)	NR	67 vs 66 NS	64.3 vs 65.5 NS	97 vs 95 NS
Axillary recurrence (%)	0 vs 1.8	1 vs 3	1 vs 3 P = 0.04	0.5 vs 1.5 P = 0.3
Ipsilateral breast recurrence (%)	1.7 vs 1.6	4 vs 2	17.2 vs 16.3	0.5 vs 0.5
Metastases (%)	5.5 vs 5.3	12 vs 10	25.8 vs 24.9	1.4 vs 3.7

Axillary Dissection Versus Observation

Martelli et al (Istituto Nazionale Tumori) randomized 219 patients between the ages of 65 and 80 with mammographic tumor size ≤2 cm to axillary dissection versus observation after quadrantectomy.4 All patients received radiotherapy to the breast (no axillary or supraclavicular radiotherapy), and all were prescribed systemic therapy with tamoxifen. There were more estrogen receptor positive (ER+)/progesterone receptor positive (PR+) patients and patients with tumors ≤1 cm in the no dissection arm, 74% versus 62% and 45% versus 30%, respectively. In the dissection arm, 23% of patients were found to have metastatic axillary lymph nodes; 74%, 2%, and 24% of lymph node positive patients had 1, 2-3, and >3 positive lymph nodes, respectively. With a median follow-up of 5 years, only 2 patients (1.8%) in the no dissection arm developed clinically evident axillary disease. There was no difference in breast cancer specific mortality (4% in each arm), rate of distant metastases (5.5% vs 5.3%), and overall survival (87% vs 92%, P = .25).

The International Breast Cancer Study Group (IBCSG) trial was similar in design in that it was a randomized comparison of axillary dissection versus observation in patients older than 60 years.5 The trial enrolled 473 patients. Fifty-five percent received breast conserving surgery, 60% percent of whom received breast radiotherapy, and all patients were prescribed tamoxifen. Recommended radiotherapy was to the breast only. More patients in the surgery arm were ER+ (84% vs 76%), and equal numbers of patients in both arms received radiotherapy. The median number of examined lymph nodes was 13, and 28% of patients were lymph node positive at the time of dissection; of these, 72% had 1-3 positive nodes, and 28% had >3 positive lymph nodes. With a median follow-up of 6.6 years, 3% of patients in the no dissection arm and 1% in the dissection arm experienced an axillary recurrence. There was no difference seen in overall survival (75% vs 73%) or disease-free survival (67% vs 66%). Because of slower than anticipated accrual for this study, the primary goal was shifted from survival to a quality of life endpoint. Despite this, power calculations done by the authors suggested that, with these results, a fully accrued trial would have had a less than 3% chance of finding a statistically significant disease-free survival benefit for axillary lymph node dissection.

Axillary Dissection Versus Axillary Radiotherapy

Updated in 2004, the trial by Louis-Sylvestre et al (Institut Curie) has the longest published follow-up in this review, with 5-, 10-, and 15-year results.6, 8 The trial included 658 patients younger than 70 years of age with tumor size <3 cm who were randomized after wide local excision to axillary lymph node dissection or definitive axillary radiotherapy. The median age of patients in this study was 52 years, younger than the prior 2 studies, and patients ≤35 years of age comprised 3% and 8% of the dissection and no dissection arms, respectively. Chemotherapy was received by 19 and 9 patients and hormonal therapy was given to 14 and 8 patients in the dissection and no dissection arms, respectively. Radiotherapy was administered comprehensively to the breast, axilla, and internal mammary lymph nodes in patients who were randomized to the axillary radiation arm. Patients with positive axillary lymph nodes in the dissection arm in addition received adjuvant supraclavicular and internal mammary lymph node irradiation. Among the enrolled patients, 21% were found to be pathologically lymph node positive in the dissection arm; 57%, 34%, and 9% had 1, 2-3, and >3 positive lymph nodes, respectively. Axillary recurrence rates were 0.6% versus 2.2% at 5 years and remained unchanged at 1% versus 3% (P = .04) at both 10- and 15-year follow-up in patients with and without axillary dissection, respectively. Overall survival had initially been found to be significantly superior in patients who underwent axillary dissection (96.6% vs 92.4%, P = .009) at 5-year follow-up. However, in the update presented by Louis-Sylvestre et al with 15-year data, the survival curves had converged with no significant difference seen in overall survival rates (75.5% and 73.8% with and without dissection, P nonsignificant). The early survival difference at 5 years was attributed by the authors to a difference in systemic therapy allocation between the randomization groups: 19 versus 9 patients received chemotherapy and 14 versus 8 patients received hormonal therapy in the dissection arm compared with the radiotherapy arm, respectively.

Axillary Radiotherapy Versus Observation

Veronesi et al (Italian Oncology Senology Group) enrolled 435 clinically lymph node negative patients older than 45 years old with tumor size <1.2 cm who underwent wide local excision or quandrantectomy.7 Patients were randomized to comprehensive radiotherapy (axilla and breast) or no axillary treatment (radiotherapy to breast only). Patient and tumor characteristics were well balanced in both arms. ER+ patients were prescribed tamoxifen, whereas patients who were ER−, had high tumor grade, and high proliferation rate, received chemotherapy. More patients received adjuvant therapy in the no axillary treatment arm, 87% versus 80%, P = .046. With 5.3 years median follow-up, the ipsilateral axillary failure rate was 0.5% in the axillary radiotherapy arm and 1.5% in the no treatment arm, P = .30. Overall survival was significantly higher in the axillary radiotherapy arm (99% vs 94%, P < .01), but this finding was attributable to an excess of death from nonbreast cancer causes in the arm receiving no radiation. Breast cancer specific death was similar between the arms: 1% versus 2.3% (P = .23), with and without axillary radiotherapy.

Meta-analysis

Meta-analysis of the 3 studies of axillary dissection versus no dissection (Martelli et al,4 IBCSG,5 Louis-Sylvestre et al6) showed no significant difference in overall survival between the 2 groups (OR 1.20; 95% confidence interval [CI], 0.64-2.27; P = .57; Fig. 1). Similarly, no significant difference was found in ipsilateral breast tumor recurrence (OR 1.12; 95% CI, 0.68-1.86; P = .66), distant metastases (OR 0.97; 95% CI, 0.69-1.38; P = .88), or disease-free survival (OR 1.08; 95% CI, 0.82-1.42; P = .61) between the 2 groups; Fig. 2). The only endpoint found to be significantly different was the rate of axillary recurrence with a greater number of failures in patients without axillary dissection (OR 0.27; 95% CI, 0.10-0.78; P = .02; Fig. 3).

Figure 1. Meta-analysis of overall survival. Odds ratio of overall survival comparing axillary dissection (treatment) to no axillary dissection (control). The 5-year data from the Louis-Sylvestre trial were used in this meta-analysis. IBCSG, the International Breast Cancer Study Group; CI, confidence interval.

Figure 2. Interconnected odds ratios of disease-free survival from individual trials. The 3 axillary interventions are placed at the apices of a triangle. Each line of the triangle represents the direct comparison between the 2 interventions at the ends of the line. All triangle arms display the same odds ratio scale. The individual odds ratios of each trial (including all 3 time points for Louis-Sylvestre) are noted. IBCSG, the International Breast Cancer Study Group.

Figure 3. Meta-analysis of axillary recurrence. Odds ratio of axillary recurrence comparing axillary dissection (treatment) to no axillary dissection (control). IBCSG, the International Breast Cancer Study Group; CI, confidence interval.

Given the results of the NSABP B-041 and Institut Curie6 trials, neither of which showed a survival difference between axillary radiotherapy and axillary dissection, a second meta-analysis was conducted with inclusion of the trial by Veronesi et al7 of axillary radiotherapy (presumed equivalent to dissection) versus no axillary treatment. The addition of the Veronesi trial to the meta-analysis did not alter the survival and recurrence endpoint results reported above. There continued to be no significant difference in overall survival with an OR of 1.55 (95% CI, 0.74-3.24; P = .24; Fig. 4). No significant difference was seen in the rates of ipsilateral local breast cancer recurrence (OR 1.11; 95% CI, 0.68-1.83; P = .67) or distant metastases (OR 0.91; 95% CI, 0.65-1.29; P = .60). Axillary lymph node recurrence continued to be the only endpoint that was significantly different between the 2 groups (OR 0.28; 95% CI, 0.11-0.73; P < .01).

Figure 4. Interconnected odds ratios of overall survival from individual trials. The 3 axillary interventions are placed at the apices of a triangle. Each line of the triangle represents the direct comparison between the 2 interventions at the ends of the line. All triangle arms display the same odds ratio scale. The individual odds ratios of each trial (including all 3 time points for Louis-Sylvestre) are noted. IBCSG, the International Breast Cancer Study Group.

DISCUSSION

This review of the literature and meta-analysis found that the only statistically significant outcome difference seen with axillary lymph node dissection in clinically lymph node negative breast cancer patients was a slight increased risk of ipsilateral axillary lymph node failure, which did not translate to an overall survival decrement in the meta-analysis.

Taken individually, the trials by Veronesi et al7 and the Institut Curie (5-year results, Cabanes et al8) both reported a higher overall survival rate in patients who underwent definitive axillary treatment. In the trial by Veronesi et al, this survival difference was attributable to an excess number of nonbreast cancer deaths in the no treatment arm (7 patients vs 0 patients, P ≤ .01), with no significant difference seen in breast cancer specific mortality. In addition, the 15-year Institut Curie update by Louis-Sylvestre et al found that the initially reported survival advantage at 5 years disappeared with longer follow-up, with no survival difference detected at 15 years.6 This initial survival advantage was attributed by the authors to an increased number of patients treated with adjuvant systemic therapy in the lymph node dissection arm, an inherent bias in the study with regard to lymph node positive patients. To preserve the homogeneity of the follow-up duration for our meta-analysis, the 5-year results from the Institut Curie trial, rather than the 15-year results, were included in the analysis, with the caveat that there was no survival difference with longer follow-up. Even given the inclusion of the earlier results, our meta-analysis did not find a survival advantage to axillary dissection, in contrast to the prior published meta-analysis by Orr.2

An increased risk of axillary failure was found in the patients who did not receive axillary dissection (or comprehensive axillary radiotherapy) in our meta-analysis as well as in each individual trial. However, the actual number of lymph node recurrences in patients not undergoing axillary dissection is much lower than would have been anticipated based on historical data. NSABP B-04 found that, although 40% of the patients who underwent dissection had pathologically positive lymph nodes, only 18% of patients who did not undergo dissection (or axillary radiotherapy) experienced a clinical axillary failure.1 This would suggest that, in our review, the rate of axillary failure in patients who did not undergo axillary treatment (1.8%, 2.2%, 3%, and 1.5%) should have been approximately 50% of the proportion of patients found to be lymph node positive at the time of dissection. However, this did not occur. Martelli et al found 23% of patients with lymph node metastases in the dissection arm, yet only 1.8% of patients clinically failed in the no dissection arm.4 Similarly, 21% of patients had axillary lymph node metastases in the trial by Louis-Sylvestre et al, but only 2.2% failed in the axilla without dissection.6 The IBCSG had a 3% rate of overt axillary failure without dissection, whereas 28% of patients had lymph node metastases in the dissection arm.5 Although the trial by Veronesi et al did not have a dissection arm, based on the patient characteristics in the trial, it was anticipated that 24% of patients were likely to have had lymph node disease.7 However, only 1.5% of patients without axillary radiotherapy had an axillary recurrence.

This discrepancy between the anticipated number and actual number of axillary failures without definitive axillary treatment is biologically interesting and may partly be attributable to the adjuvant therapy received by patients in these studies. In contrast to NSABP B-04,1 the majority of patients in our review4-7 received adjuvant systemic therapy (hormonal therapy or chemotherapy) and received whole breast irradiation. A dosimetric study by Reznik et al reported that, with standard whole breast tangential fields, the level I and II axilla receive on average 66% and 44%, respectively, of the prescribed dose.9 In addition, Wong et al found no isolated axillary failures after tangential breast radiotherapy alone, without axillary dissection, in a select group of 92 early stage clinically lymph node negative breast cancer patients.10 The anticipated rate of lymph node metastases in this group was approximately 24%. Greater than half of their patients received systemic therapy, and some of the patients in that study were treated with “high” tangential radiation fields, which treat more of the axilla than conventional tangential radiation portals.

It should be emphasized that the results found in this study contradict the Orr meta-analysis in which a 5.4% overall survival benefit was described in patients who underwent axillary dissection.2 This finding might in part be explained by the lower risk population that was enrolled in the studies included in our meta-analysis. More rigid enrollment criteria were enacted by the trials in our review, excluding younger patients with larger tumors. The participation in mammographic screening also inherently biased the entire study population toward presentation with earlier stage disease, thereby decreasing the overall risk of subclinical lymph node involvement.11 Also of importance is the much greater use of systemic therapy and increased number of patients who received breast radiotherapy (after breast conservation) in the more recent trials included in our review; this adjuvant therapy might account for a lower risk of axillary failure overall.

The omission of axillary dissection in the studies reviewed did lead to a lack of staging information for the approximately 20%-30% of patients who would likely have been found to be lymph node positive if surgically evaluated. Because this staging information holds prognostic implications for survival, with emerging data that even patients with micrometastatic disease may have a worse survival than lymph node negative patients,12 an issue of interest is whether this pathological information would have impacted treatment decisions in those patients. Specifically, would more or different systemic therapy have been offered to those patients had their pathological lymph node information been available? Most likely, in cases of micrometastatic lymph node involvement, only hormonal therapy would have been offered, and the majority of patients in these studies received hormonal therapy despite uncertainty about the pathological status of the axilla. In addition, because there was no survival decrement in patients who did not receive dissection, an inference that can be made is that the systemic treatment they received was adequate and equivalent to the patients in the control arm of the trials.

The adoption of sentinel lymph node biopsy into clinical practice has significantly decreased the number of patients undergoing full axillary lymph node dissection. Sentinel lymph node biopsy lends important pathological information with a procedure that is minimally morbid. The results of this study suggest that, in lower risk patients, even sentinel lymph node biopsy could likely be safely withheld as the pathological information is unlikely to change their adjuvant therapy options and their overall survival is not likely to be impacted. Lower risk patients that could qualify for such an approach would be older patients with small, hormone receptor positive tumors, who are clinically lymph node negative, and will be receiving adjuvant hormonal suppression and whole breast radiotherapy. In women who are estrogen receptor negative, information from sentinel lymph node biopsy may be helpful in guiding the decision regarding chemotherapy.

Our meta-analysis is limited by a small number of studies and low number of overall events. Although the patient and disease characteristics were fairly similar across all of the trials, there was variability between the randomization arms in the trials, most notably in the trial by Veronesi et al, which did not include an axillary dissection arm. The results of this meta-analysis should be interpreted with the understanding of this design. In addition, these findings may or may not be generalizable to patients with more advanced breast cancer, by virtue of large tumor size or clinically positive lymph nodes.

In summary, this review and meta-analysis of contemporary randomized trials of axillary lymph node treatment demonstrates that no survival benefit is associated with axillary lymph node dissection in patients with early stage clinically lymph node negative breast cancer. A small increase in axillary failure rate is evident in the absence of definitive axillary treatment, but the absolute number of axillary failures is extremely low in the setting of adjuvant systemic therapy and breast radiotherapy.

Conflict of Interest Disclosures:

The authors made no disclosures.