Pooled Long-Term Outcomes from Two Randomised Trials of Axillary Node Sampling with Axillary

Pooled long-term outcomes from two randomised trials of axillary node sampling with axillary radiotherapy versus axillary node clearance in patients with operable node positive breast cancer

Bing AU Edinburgh Breast Unit, UK, Kerr GR Oncology Department, Western General Hospital, Edinburgh, UK, Jack W Edinburgh Breast Unit, UK, Chetty U Edinburgh Breast Unit, UK, Williams L Centre for Population Health Sciences, University of Edinburgh, UK, Rodger A Edinburgh Breast Unit, UK, Dixon JM Breakthrough Research Unit and Edinburgh Breast Unit, University of Edinburgh, UK.

Correspondence to: Alison Bing

Tel. 0131 537 2643

Fax. 0131 537 2653

No funding to declare

Category: Original article

This research was presented as a poster at the Miami Breast Cancer Conference 26/02/2015.

This research was presented as an oral presentation at the Association of Breast Surgery Conference and AGM, Bournemouth 15/06/2015.

Key words and phrases: Axillary node clearance, axillary node sample, sentinel lymph node biopsy, breast cancer, axillary radiotherapy.

Abbreviations: ANC = axillary node clearance, ANS = axillary node sample, AXRT = axillary external beam radiotherapy, TNM = tumour nodes metastasis classification, chemo = chemotherapy, NHS =National Health Service, SLNB = sentinel lymph node biopsy, ER= estrogen receptor, CI = Confidence Interval, NICE = National Institute of clinical Excellence, ASCO = American Society of Clinical Oncology, RCT = Randomised Controlled Trial, CMF = Cyclophosphamide, methotrexate and fluorouracil.

Abstract

Objectives: To determine long-term overall, breast cancer specific and metastasis-free survival as well as axillary relapse rate from a pooled analysis of two randomised trials in women with operable breast cancer. These trials compared axillary node sampling (ANS) combined with axillary radiotherapy (AXRT), if the sampled nodes were involved, with axillary node clearance (ANC).

Methods: In total, 855 patients were included in the trials. Out of these, 62.5% were node negative and randomised to ANS vs. ANC. The node positive patients (37.5%) were further treated with AXRT, thus allowing for analysis of ANS+AXRT vs. ANC. Follow-up to death or 20 years was available in 799 women with a median follow-up of 19.4 years and mean follow-up of 19.9 years.

Results: There was no evidence of a breast cancer survival advantage for ANS vs. ANC in node negative patients; HR 0.88 (95% CI 0.58-1.34, p=0.55) or ANS + AXRT vs. ANC in node positive patients; HR 1.07 (95% CI 0.77-1.50, p=0.69). There was no metastasis-free survival advantage for ANC vs. ANS+/-AXRT in node negative; HR 1.03 (95% CI 0.70-1.51, p=0.88) or node positive patients; HR 1.03 (95% CI 0.75-1.43, p=0.85). There were more axillary recurrences in node negative patients having ANS vs. ANC; HR 3.53 (95% CI 1.29-9.63, p=0.014) and node positive patients having ANS +/- AXRT vs. ANC; HR 2.64 (95% CI 1.00-6.95, p=0.05).

Conclusions: Despite a higher rate of axillary recurrences with axillary node sampling combined with radiotherapy to the axilla, an axillary node clearance did not improve overall, breast cancer specific or metastasis-free survival. Axillary recurrence is thus not a satisfactory endpoint when comparing axillary treatments.

Introduction

Management of the axilla in invasive breast cancer remains an area of on-going controversy and debate. For many years it was practice for patients with invasive cancer to have an axillary lymph node clearance (ANC) regardless of whether nodes were thought to be involved pre-operatively (1, 2, 3).

ANC carries considerable morbidity in terms of lymphoedema, pain, damage to the intercosto-brachial nerve, and interference with shoulder movement (4, 5, 6). Axillary node sampling (ANS) is a technique pioneered in the Edinburgh Breast Unit in the early 1980s. The aim was to offer a minimally invasive operative approach to assess the status of the axillary nodes and avoid ANC in patients without histological evidence of axillary node involvement. It was a technique that others did not find easy to learn and was therefore not widely used outside of a few major Breast Units. The ANS technique was developed and validated through a series of randomised trials and was shown to provide a minimally invasive accurate method of assessing axillary nodal status (7, 8, 9). ANS was superseded first by blue dye directed for node axillary sampling and then by sentinel lymph node biopsy using both radioisotope and blue dye (10, 11). The ALMANAC trial (6) showed less arm morbidity and better quality of life for SLNB when compared with ANC, with similar morbidity for both ANS and SLNB. The reduction in morbidity is the major reason why SLNB is now widely used in clinically node negative patients.

The move to perform fewer axillary clearance operations has been shown to be safe at least in the first decade after treatment with equivalent survival and local recurrence rates with SLNB when compared to ANC for women without axillary node involvement. Even in node positive patients the routine use of ANC is now being questioned. In patients with one or two involved nodes on SLNB, having breast conserving surgery and whole breast radiotherapy, the ACOSOG Z0011 trial showed no improvement in axillary recurrence rates or overall survival for ANC compared with SLNB alone (12). Recently the AMAROS (13) trial has shown that in patients with involved nodes on SLNB, AXRT gives equal short-term outcomes to ANC.

Although the National Institute of Clinical Evidence (NICE), still recommends ANC for axillary lymph node positive breast patients (14), American Society of Clinical Oncology (ASCO) guidelines state that patients with one to two positive nodes having breast conserving surgery and whole breast radiotherapy who fulfil Z0011 entry criteria do not require routine axillary lymph node clearance.

Over thirty years ago the Edinburgh Breast Unit initiated two randomised controlled trials (RCT) (7, 8) comparing axillary lymph node sampling followed by axillary radiotherapy if the sampled nodes were involved with axillary node clearance. Here updated results from these two cohorts (7, 8) are presented, and provides the first long term data comparing ANS+/- AXRT with ANC on overall survival, breast cancer specific survival, metastasis-free survival and axillary recurrence patterns of patients with a follow- up of up of over 20 years from treatment.

Materials and Methods

Trials included in the pooled analysis

Two consecutive RCTs on ANS versus ANC were used in the pooled analysis; Management of the axilla in operable breast cancer treated by breast conservation: a randomised clinical trial (Chetty U et al) (7) and The Edinburgh randomized trial of axillary sampling or clearance after mastectomy (Forrest et al) (8). analysis. The first enrolled patients having mastectomy and the second included patients treated by breast conserving surgery. Details of studies are found below.

Approval was sought for both trials from a local ethics committee, and informed consent was taken from all patients.

The inclusion criteria for the two trials were: operable invasive breast cancer (T1-T3, N0-N1, M0), the patient being suitable for both surgery and radiotherapy. The exclusion criteria included: inoperable or clinically multi-centric tumour, ductal carcinoma in situ, fixed axillary nodes, previous invasive carcinoma (except skin basal cell carcinoma), Paget’s disease of the nipple and male gender. In the conservation trial patients over the age of 70 years were also excluded (7). All patients underwent investigation to exclude metastatic disease including liver ultrasound scan, skeletal scintigraphy, radiography of the chest, abdomen and pelvis, haematological and liver function testing. The study endpoints were: overall survival, breast cancer specific survival, time to loco-regional relapse and time to distant metastases.

Surgery of the axilla

In ANS the surgeon identified four palpable axillary lymph nodes. These nodes were primarily taken from level I in the lower axilla and the number of nodes excised ranged from 2-8. Patients randomised to ANC had all nodes removed up to and including level III. Node involvement was determined histologically following ANS or ANC.

Radiotherapy of the axilla

Patients with node positive disease found after ANS were treated with AXRT. AXRT was given to all node-positive sampled patients apart from 5 patients who were randomized to the 'no radiotherapy' arm of the Scottish Conservation Trial, and these patients have been omitted from the analysis in the present study. Patients who underwent ANC and who had involved nodes did not receive radiotherapy.

In the breast conservation trial regional lymphatics were treated by ‘a direct anterior field covering the axilla and supraclavicular fossa with a posterior axillary boost bringing the mid-axillary dose to 4500Gy given in 20 fractions over 4 weeks’ (7).

In the mastectomy trial radiotherapy included ‘the ipsilateral internal mammary chain, shoulder fields and supraclavicular fossa’ (8). During the course of the trial modifications were made to the radiotherapy treatment with the total dose being reduced from 4500Gy in 20 fractions to 4000Gy given in 10 fractions over 4 weeks.

Systemic Therapy

All patients received standard post-operative adjuvant chemotherapy or endocrine treatment for the time period of the study. Chemotherapy was mainly given by the cyclophosphamide, methotrexate and fluorouracil (CMF) regimen or with adriamycin-based regimens. Treatments were based on the results of pathology, including axillary lymph node histology and hormonal receptor status. Some ER negative patients did not get chemotherapy and many ER negative patients received tamoxifen because of the belief at the time that also these patients might benefit from this treatment. In addition, not all women with ER positive cancers got tamoxifen. Some pre-menopausal women with ER positive cancers had an oophorectomy.

Patient follow-up

Patients were followed in outpatient clinics in the oncology and breast surgery departments, 3 monthly for 2 years, 6 monthly from years 2-5 and annually thereafter. Patients who were disease free at 10 years were discharged to the care of their primary care doctor. Follow-up in these patients was by yearly contact with the general practitioner.

Data available from the pooled trials

The demographic information available included: age, type of breast cancer, TNM stage, tumour size on pathology and lymph node status (number of nodes excised, number of positive nodes), type of axillary operation, radiotherapy to the axilla, radiotherapy to the breast, oestrogen receptor status, adjuvant systemic therapy given.

Data were updated from previous publications using existing departmental databases, electronic patient records and case notes. Cause of death was established from death certificates archived at Register House, Edinburgh. Breast cancer was only considered to be the cause of death if listed within part 1 of the death certificate.

Data analysis

Data analysis was carried out using SPSS software (IBM SPSS Statistics for Windows, Version 19.0. Armonk, NY: IBM Corp.) by an independent statistician (LW). Kaplan-Meir survival curves were used to show survival and relapse rates. Cox survival models were used to generate hazard ratios and 95% confidence intervals. The proportional hazards assumption was tested and found to hold in all analyses.

The log rank test was used to test for equality of survival distributions and to assess any statistically significant differences between randomised treatment groups. All tests were two-sided. A limited number of subgroup analyses were performed to generate hypotheses.

Results

Long-term data available for analysis

In total, 855 patients with invasive breast cancer were randomized to either ANC (n=424) or ANS +/- AXRT (n=431) (figure 1) within the two trials between 1980 and 1995. 799 patients had follow up to death or at least 20 years, with only 56 patients (6.5%) being lost to follow up, in the most part due to change of address. The median and mean follow-up times were 19.4 and 19.9 years, respectively.

Node positive patients

Information was available on 301 node positive patients (37.7% of all patients). This excludes patients randomised to the ‘no radiotherapy’ arm of the Scottish Conservation Trial and patients where the number of nodes was not recorded. This node positive group was subdivided into patients with 1, 2-4, 5-9 and ≥10 positive nodes (table 1).

Overall survival

The overall median survival time in these patients was estimated at 13.8 years (ANC 13.2 years and ANS + AXRT 14.4 years). The log rank test showed no difference in overall survival between ANC and ANS (p=0.88).

There was no evidence to suggest that, in node positive patients, ANS +AXRT gave a survival advantage or disadvantage over ANC, HR 0.98 (95% CI 0.75-1.28, p=0.88).

There was no significant difference in survival between patients with 1 node and 2-4 nodes involved (p=0.16). However, patients with 5-9 (HR 1.73; 95% CI 1.11-2.71, p=0.016) and ≥10 involved nodes (HR 2.45; 95% CI 1.51- 3.98 p=<0.001) had significantly higher risks of death than patients with only one node involved.

In terms of overall survival, age over 65 years (HR 2.18; 95% CI 1.51-3.15) and number of involved axillary nodes (THIS COULD BE REMOVED COMPLETELY AS REPEATING THE ABOVE PARAGRAPH >10 involved nodes, (HR 2.45; 95% CI 1.51- 3.98 p=<0.001))HR 2.56; 95% CI 1.58-4.16) were the only significant predictors of time to death. Axillary surgery was not found to be a significant predictor of death once other factors had been accounted for in node positive patients (p=0.25).

Breast cancer specific survival

In axillary lymph node positive patients, 137 of the 221 deaths were due to breast cancer (62.3%). There was no statistically significant evidence to suggest that, in lymph node positive patients, ANS+AXRT gave a survival advantage or disadvantage over ANC for death from breast cancer, HR 1.07 (95% CI 0.77-1.50). The log rank test showed no difference in breast cancer deaths between ANC and ANS groups (p=0.69).

Axillary relapse

In node positive patients, a difference in time to axillary recurrence between ANC and ANS + AXRT was shown (p=0.04). Patients who had ANS + AXRT were more than twice as likely compared with those having ANC to develop an axillary recurrence, HR 2.64 (95% CI 1.00-6.95 p=0.05). Absolute numbers of axillary recurrences, by type of axillary surgery and node status, are detailed in table 2.

Time to metastases

There was no evidence to suggest that ANS+AXRT had a metastasis-free survival advantage or disadvantage over ANC (HR 1.03; 95% CI (0.75-1.43), p=0.85). The log rank test revealed no evidence to suggest a difference in time to metastatic disease, between ANC and ANS+AXRT (p=0.85) for node positive patients.

As expected, patients with a greater number of positive nodes had a significantly higher risk of metastasis. Patients with 5-9 nodes vs. less involved nodes had a HR 2.24 (95% CI 1.33-3.76, p=0.002), and patients with >10 nodes vs. less involved nodes having a HR 3.45 (95% CI; 1.99-5.97, p=<0.001).