Axillary Dissection

Hiram S. Cody III

George Plitas

INTRODUCTION

For patients with operable breast cancer, axillary node status remains the single most important prognostic factor. The primary goal of axillary surgery is staging to govern the use and type of systemic therapy; secondary goals include local control and the possibility of a small survival benefit. Axillary lymph node dissection (ALND) has been regarded for most of the 20th century as the “gold standard” operation to achieve these goals, but has largely been replaced over the past decade by sentinel lymph node (SLN) biopsy, first reported by Krag et al. in 1993 (1) and Giuliano et al. in 1994 (2). SLN biopsy allows the avoidance of ALND in SLN-negative patients, sparing them the morbidity of a larger operation, and allows the routine performance of additional pathologic studies, potentially increasing the accuracy of staging. Sixty-nine observational series (3) of SLN biopsy validated by a “backup” ALND, three meta-analyses (4, 5 and 6), and the early results of seven randomized trials (4) comparing ALND with SLN biopsy confirm that the morbidity of SLN biopsy is less than ALND, that staging accuracy is at least equivalent, and (in the single randomized trial reporting long-term results (7) that survival and other disease-related adverse events are comparable at 7 years’ follow-up. While the role of ALND in the era of SLN biopsy has been reduced, it has not been eliminated, and this chapter surveys the historic evolution, current indications, operative technique, and morbidity of ALND. Looking ahead, prognostication through rapidly emerging genomic technologies may eventually rival (or even surpass) that of conventional histopathology, and the role of ALND will continue to change.

THE HISTORIC EVOLUTION OF ALND

Jean Louis Petit (1674-1750), director of the French Surgical Academy, was probably the first surgeon to articulate a unified concept for breast cancer surgery (8). He emphasized the importance of an en bloc resection of the breast and axillary nodes, but his insight came too early: even by the mid-19th century, breast cancer was widely regarded as incurable by surgery. Halsted’s landmark 1894 (9) and 1907 (10) reports of his meticulous technique for “radical mastectomy” (RM, which included removal of the breast and pectoral muscles, with a complete ALND) were the first to demonstrate that coincident with a striking reduction in LR (from 51% to 82% reported by European center to 6%), 31% of patients (a significant proportion at that time) were disease-free at 5 years. This intuitive concept relating local control and survival made RM the standard operation for breast cancer over the next 70 years, despite subsequent reports of techniques that were less radical, including modified radical mastectomy (MRM) (11). In the “Halstedian” era, the goal for ALND (as for mastectomy) was to maximize cure by minimizing local failure.

Coincident with the acceptance of MRM in the 1970s, Fisher (12) proposed that breast cancer survival was largely a function of tumor biology and not surgical technique. The “Fisher hypothesis” was tested in National Surgical Adjuvant Breast and Bowel Project (NSABP) B-04 (1971-1974) (13); patients with clinically node-positive breast cancer were randomized to RM versus total mastectomy/radiotherapy (RT), and patients with clinically node-negative breast cancer were randomized to RM versus total mastectomy/RT
versus total mastectomy alone. At 25 years’ follow-up there were no differences in any category of survival (overall, disease-free, distant disease-free) between the patients in the two node-positive arms, or in the three node-negative arms of the trial. B-04 confirmed the overwhelming prognostic significance of axillary node metastasis and for this reason, ALND was incorporated into all subsequent NSABP trials for invasive breast cancer.

With a series of remarkable meta-analyses from the Early Breast Cancer Trialists Collaborative Group (EBCTCG), it has become clear that breast cancer is best viewed as a disease with a wide spectrum of behavior (14), rather than a predominantly local (Halsted) or systemic (Fisher) process. Separate EBCTCG overviews show that local control and survival are related (15) but that there is no survival advantage for more radical versus less radical versions of mastectomy (or for mastectomy vs. breast conservation) (16), and that there is an incremental survival benefit from the addition of systemic adjuvant therapy to local treatment (17). At present, virtually all node-negative patients are staged by SLN biopsy alone and the principal goal of ALND is to maximize local control in patients already proven by SLN biopsy to be node-positive.

ALND VERSUS OTHER METHODS OF STAGING

ALND can be compared with other methods of axillary staging. These are (i) no axillary surgery (with or without axillary RT), (ii) axillary sampling, and (iii) SLN biopsy.

ALND versus No Axillary Surgery

The foremost trial comparing ALND with no axillary surgery is NSABP B-04 (13) (Table 38-1), as described above. Among patients randomized to total mastectomy alone, 18% developed axillary LR as the first sign of treatment failure and required a delayed ALND; 79% of axillary LR occurred within 2 years and 95% within 5 years. Two more recent trials demonstrate far lower rates of axillary LR in older patients treated without ALND. Martelli et al. (18) randomized 219 patients (aged 65 to 80) with T1N0 disease to breast-conservation surgery with or without ALND, and all patients received 5 years of tamoxifen. At 5 years’ follow-up there were no differences in disease-free or overall survival, and axillary LR in the no-ALND arm was 1%. Rudenstam et al. (19) randomized 473 patients (aged 60 years or more) to breast surgery with or without ALND; all patients received 5 years of tamoxifen. At 6.6 years’ follow-up there were no differences in disease-free or overall survival, and axillary LR in the ALND and no-ALND arms was 1% and 3%, respectively.

The addition of axillary RT improves local control in patients treated without axillary surgery (Table 38-1). In NSABP B-04 (13), locoregional recurrence at 10 years was lower with total mastectomy/RT than with total mastectomy alone (5% vs. 31%), as was axillary LR (3% vs. 19%). In the Cancer Research Campaign (King’s/Cambridge) trial (20), 2,268 patients were randomized to total mastectomy/RT (to chest wall and axillary nodes) versus total mastectomy alone; again, crude LR was lower in the RT group (5% vs. 15%), as was axillary LR (2% vs. 13%). In a randomized trial from the Institut Curie, the authors compared the results of ALND and axillary RT in 658 patients; they observed a survival advantage for ALND at 5 years (21), but no survival differences between groups at 10 and 15 years (22). Axillary LR occurred slightly less often after ALND than after axillary RT (1% vs. 3%, p = .04). Finally, Veronesi et al. (23) randomized 435 patients, none of whom had ALND, to breast conservation with or without axillary RT. At 5 years’ follow-up they found no differences in disease-free survival, and axillary LR in the axillary RT and no-ALND arms was 0.5% and 1.5%, respectively. Three observational studies also report high rates of axillary LR in the untreated axilla and also show that axillary LR is highly dependent on tumor size (Table 38-1) (24, 25 and 26). Tumor characteristics alone, however, cannot reliably predict axillary node status with greater that 90% to 95% accuracy (27). Bevilacqua et al. (28) have recently developed a multivariate nomogram for the prediction of SLN metastases, using a sophisticated model based on 3,786 SLN biopsy procedures and prospectively validated in 1,545 subsequent procedures. They too find that the prediction of SLN status is imperfect, with only a 75% chance, between two randomly selected individuals (one of whom is node-positive), of correctly identifying the node-positive patient.

Others have asked whether noninvasive imaging can replace surgical staging. Neither CT nor MRI is adequate for lymph node staging. PET lacks the resolution to detect metastases smaller than 5 mm, so is subject to false-negative and false-positive results; in five reports, sensitivity ranges from 27% to 94%, and specificity from 43% to 97% (29, 30, 31, 32 and 33). The results of axillary ultrasound (US) with fine-needle aspiration (FNA) vary widely, reflecting differences in methodology and case selection, but allow triage of FNA-positive patients directly to ALND (34). US-guided FNA of axillary nodes can spare patients the added time and cost of SLN biopsy, but is insufficiently sensitive to replace surgical staging.

ALND versus Axillary Sampling

As practiced in the United Kingdom, axillary sampling is a limited staging operation in which about four nodes are removed from the low axilla, guided by intraoperative palpation. Two randomized trials from Edinburgh have compared axillary sampling with ALND, for patients having mastectomy (with 11 years’ follow-up) (35) or breast conservation (wide excision/breast RT, with 4 years’ follow-up) (36) (Table 38-2). Node-positive patients in each trial received axillary RT. Between sampling and ALND, the authors observed a comparable proportion of positive axillae, comparable rates of axillary LR (5.4% vs. 3%), and comparable survival between the two arms of each study, but greater long-term shoulder morbidity for patients who had sampling/RT compared to ALND. Since none of the UK axillary sampling data have been validated by a backup ALND (as has been done for SLN biopsy), one cannot calculate the performance characteristics of this method. In a separate Swedish study by Ahlgen et al. (37), axillary sampling (“five node biopsy”) was validated by a planned backup ALND in 415 patients, and sensitivity for cN0 patients was 95.5%.

ALND versus SLN Biopsy

Seven randomized trials compare ALND and SLN biopsy (4), allocating patients to ALND versus SLN biopsy (plus ALND for SLN-positive patients), and collectively confirm that the staging accuracy of ALND and SLN biopsy is comparable, and that the morbidity of SLN biopsy is less. For two of the trials (38, 39) patients in the ALND arm also had SLN biopsy, confirming false-negative rates for SLN biopsy of 8.8% and 9.7%, respectively. In the one trial (7) reporting long-term follow-up there were no differences in survival or in any other disease-related adverse events at 7 years, and there was a single case of axillary node recurrence following a negative SLN biopsy.

TABLE 38-1 Studies of Axillary Treatment (ALND or RT) versus No Axillary Treatment in cN0 Breast Cancer

	NSABP B-04 (13) (1971-1973)	Milan (18) (1996-2000)	SBCSG (19) (1993-2002)	King’s/Cambridge (20) (1970-1975)	Curie (22) (1982-1987)	EIO (23) (1995-1998)	Baum and Coyle (26) (1973-1977)	Baxter et al. (25) (1977-1986)	Greco et al. (24) (1986-1994)
Design	RCT	RCT	RCT	RCT	RCT	RCT	Cohort	Cohort	Cohort
Breast treatment	Mastectomy	BCT	BCT or mastectomy	Mastectomy	Wide excision/RT	BCT	Mastectomy	Wide excision/RT	Wide excision/RT^a
Axillary treatment	ALND vs. Ax RT vs. none	ALND vs. none	ALND vs. none	Ax RT vs. none	ALND vs. Ax RT	no ALND vs. Ax RT	none	none	none
No. of patients	1,079	219	473	2,268	658	435	48	112	401
Follow-up	25 yr	5 yr	6.6 yr	1-5 yr	15 yr	5 yr	1-4 yr	10 yr	5 yr
Axillary local recurrence	ALND 1.4% Ax RT 3.1% none 19%	ALND 0% none 1%	ALND 1% none 3%	Ax RT 5% none 15%	ALND 1% Ax RT 3%	none 1.5% Ax RT 0.5%	21%	T1a, b 9% T1c 26% T2 33% Overall 28%	T1a 2.0% T1b 1.7% T1c 10% T2 18% Overall 7%
Overall survival	no difference	no difference	no difference	no difference	no difference	no difference	—	—	—
NED survival	no difference	no difference	no difference	no difference	no difference	no difference	—	—	—
SBCSG, Swedish Breast Cancer Study Group; EIO, European Institute of Oncology; ALND, axillary lymph node dissection; RCT, randomized controlled trial; RT, radiotherapy; NED, no evidence of disease; wide excision/RT indicates RT to breast only; ax RT indicates RT to axilla. ^a96% had wide excision/RT and 4% had mastectomy.