The accuracy of preoperative staging of the axilla in primary breast cancer: a national register based study on behalf of Danish Breast Cancer Group (DBCG)

Introduction

Staging of axillary lymph nodes in women with breast cancer is an important guide for treatment decisions. For decades, axillary lymph node dissection (ALND) was the standard procedure in staging the axilla, but today, sentinel lymph node biopsy (SLNB) is the standard of care in clinically node negative women.

In Denmark, patients suspicious of breast cancer are referred for a triple test assessment which consists of clinical examination, mammography, whole-breast ultrasonography (US) and needle biopsy of suspicious lesions. The preoperative examination also includes US of the axillary lymph nodes and fine needle aspiration cytology (FNAC) if enlarged or suspicious lymph nodes are present [1]. This examination is an important tool in the preoperative staging of patients with primary breast cancer [2,3].

Patients are classified as clinically node negative if no suspicious axillary lymph nodes are seen on US. These patients will be offered a SLNB, followed by ALND if macrometastases (tumor deposits >2 mm) are found in the sentinel lymph node(s). SLNB was completely implemented in Denmark by the end of 2004 [4].

Patients with preoperatively verified axillary metastases will have ALND performed immediately or receive neoadjuvant treatment. Accurate preoperative axillary lymph node status can reduce the numbers of patients having unnecessary SLNB performed. This reduces the time of the surgical procedure and it has been shown to lower healthcare costs [5,6].

The sensitivity of the preoperative staging of the axilla has been shown to vary between studies from [2,5]. In the meta-analysis by Diepstraten et al. [5] the sensitivity were found to be 50%.

The primary aim of this study was to evaluate the accuracy of preoperative staging of axillary lymph node involvement in patients with primary breast cancer based on preoperative US with or without FNAC. It was done in a large population-based Danish breast cancer cohort in a period with routine use of diagnostic procedures and mammographic screening.

Material and methods

The patient cohort was retrieved from the Danish Breast Cancer Group (DBCG) database. All departments in Denmark involved in the treatment of breast cancer report information regarding diagnosis, individual patient data, type of surgery, and adjuvant treatment to the DBCG database [7].

Study design and study population

The study is a retrospective register-based study which includes women diagnosed with unilateral primary breast cancer between 1 January 2010 and 31 December 2015. Only patients, in whom staging of the axillary lymph nodes was performed, were included. The exclusion criteria were patients treated with neoadjuvant therapy, patients not receiving any surgical treatment, patients with previous surgery in the breast or axillary region and patients with occult breast cancer. Furthermore, patients with inflammatory, multifocal or ulcerating breast cancer were excluded since ALND was standard of care in these patients regardless of the result of the axillary US. Patients with lymph node metastases found using other imaging methods, such as MRI and PET scanning were excluded.

More than 2500 patients were manually cross checked using the Danish Pathology Data Bank, due to missing data, and were either excluded or included according to listed criteria. See flowcart and Appendix I.

The study was approved the Danish Data Protection Agency and the Danish Patient Safety Authority.

Specificity and sensitivity

The proportion of patients who were preoperatively clinically node-negative and subsequently had macrometastases identified on SLNB were calculated. Additionally, we identified the number of patients with preoperatively verified axillary involvement on US and FNAC and afterwards confirmed on ALND. Sensitivity of US and FNAC was defined as the total number of patients with a positive preoperative US + FNAC examination among the patients with macrometastases in the axilla. Furthermore, we looked at the specificity, defined as the number of patients with a negative US and FNAC examination among the patients without macrometastases in the axilla. The positive predictive value (PPV) for predicting the risk of having axillary macrometastases with a positive US + FNAC, and the negative predictive value (NPV) for predicting the probability of not having axillary macrometastases with a negative US + FNAC was calculated. The sensitivity and specificity of preoperative axillary US and FNAC was calculated accordingly.

The difference between the mean numbers of macrometastases in the SLNB group compared to the group who had ALND performed initially was tested with a Wilcoxon rank test. p < .05 was considered statistically significant. A Chi-squared test was used to test whether histologic subtype was a risk of overseeing macrometastases on the preoperative examination. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. All statistical analyzes were performed using SAS statistical software, version 5.1 (SAS Institute Inc., Cary, NC).

Results

A total of 20,498 patients were retrieved from the DBCG database during the 5-year study period. After exclusion according to the listed criteria 18,968 patients remained for further analysis (Figure 1).

Figure 1. Flowchart 20,498 Danish womem with primary breast cancer treated 2010–2015. Macromet: Macrometastasis; LN: Lymphnode.

The prevalence of macrometastases in the axillary lymph nodes in the present study cohort was 29.1% (Table 1). Of these, 17,265 (91.0%) were clinically node negative and had SLNB performed. The remaining 1703 patients (9%) did not have SLNB performed as a consequence of a positive preoperative US and FNAC. A positive preoperative examination was defined as a suspicious US and malignant cells or cells suspicious of malignancy in the axillary lymph nodes on FNAC leading to an immediate ALND. In a few patients, ALND was performed based only on a suspicious US without FNAC.

Table 1. Distribution according to histologic subtype of 5496 Danish women with primary breast cancer and macrometastasis in axillary lymph nodes treated in the period of 2010–2015.

	Procedure	Procedure
	SLNB	ALND	Total
Diagnosis
Ductal	3126 (82.0%)	1479 (87.9%)	4605
Lobular	536 (14.1%)	100 (6.0%)	636
Other	152 (4.0%)	103 (6.1%)	255
Total	3814	1682	5496

In all, 13,469 patients (78%) did not have macrometastases in the sentinel node, but 194 of these patients had macrometastases in non-sentinel nodes, and 242 patients had non-detection of SN and had ALND performed. Out of these, 150 did not have any macrometastases. In total, 13,425 patients did not have macrometastases and were defined as true node-negative. Of the 17,265 clinically node negative patients, 3554 patients had macrometastases in the sentinel node, 194 patients had macrometastases in non-sentinel nodes and 92 patients in the non-detection group had macrometastases on the subsequent ALND (Figure 1). This resulted in 3840 (22.2%) patients who were false negative on the preoperative examination.

Out of the 1703 patients who did not have a SNLB performed, 1691 patients had macrometastases on ALND and were true positive. The remaining 21 patients did not have macrometastases on ALND, but eight of the patients had micrometastases and one patient had isolated tumor cells. These nine patients were not considered false positive. In total, 12 patients did not have any malignant cells on ALND and were defined as false positive. These 12 patients had suspicious malignant cells on FNAC (n = 10) or a suspicious US (n = 2).

The sensitivity of the combined preoperative examination including US and FNAC was 30.6%. The risk of overlooking macrometastases on the preoperative examination was associated to lobular histologic subtype compared to ductal histologic subtype, OR = 2.54 (95% CI: 2.03–3.17). There was a significant difference between the mean number of macrometastases in the group who has SLNB performed (2.8) compared to those who have ALND performed initially (5.8), p < .0001. The specificity of the combined preoperative examination was 99.9%. PPV was 99.3% based on the 1691 patients who were true positive and the total number of lymph node positive patients. Hence, the NPV was 77.8%, calculated from the number of 13,425 patients being true negative and the total number of negative (13,425 + 3840 patients).

Discussion

The implementation of SLNB is the largest breakthrough in breast cancer surgery since the introduction of breast conserving surgery in the eighties. Today, about 65% of breast cancer patients are spared an ALND without impairment of prognosis and local control, but gaining much less arm morbidity and increasing quality of life [8]. However, as long as ALND is the standard of care in patients with positive SN, the SLNB is basically, in case of macrometastatic spread, only waste of time, money and induction of false hope for the patient.

The present study indicates that only about one-third of women with macrometastases are identified prior to surgery. Not surprisingly the chance of finding metastatic lymph nodes preoperatively was associated with high metastatic burden in the axilla. During recent years, international randomized trials have indicated that that omission of ALND is safe in women with minor spread to the axillary lymph nodes [9,10]. This increases the need for a precise preoperative staging of the axilla in order to identify patients with major metastatic spread, where ALND may still be indicated.

An increasing proportion of patients with verified metastases in the axilla by axillary US and FNAC are offered neoadjuvant treatment. Around 30% of these patients experience pathologic complete response (PCR) in the axilla [11,12]. Recent studies have shown that ALND can safely be omitted in patients with PCR if no metastases or even isolated tumor cells are found in the SN or in a FNAC proven metastatic lymph node, marked before treatment [13,14]. Again, this increases the need for a more accurate preoperative staging of the axilla.

The study shows that the Danish diagnostic set-up is very predictive concerning positive axillary status since only on average two women in Denmark of the about 4700 new cases of breast cancer every year, undergo direct ALND based on a false-positive preoperative diagnosis of axillary spread. However, two-thirds of the patients with macroscopic positive lymph nodes were diagnosed on SLNB, which prolonged the surgical procedure, and excluded axillary status as indication for neoadjuvant treatment. Data from DBCG shows that the sensitivity varies from 24 to 48% between the different centers in Denmark [15]. Hence, it is possible to optimize the present technology. The need for new diagnostic tools to identify nodal status in women with breast cancer is obvious. At present a new PET-tracer is tested at Rigshospitalet with particular focus on sensitivity of axillary status.

Acknowledgment

We would like to thank Maj-Britt Jensen for statistical support.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was funded by Danish Cancer Society.

Appendix 1: Postop. axilstatus.

Reasons for exclusion of patients:

• – Neoadjuvant treatment (n = 1201)

• – No surgical treatment (n = 120)

• – No axillary dissection (n = 60)

• – Multifocal or ulcerating tumor (primary ALND) (n = 27)

• – Relapse of previous cancer or DCIS (n = 25)

• – DCIS/LCIS (n = 23)

• – Primary ALND with no reason noted (n = 16)

• – Previous surgery in breast/axil, therefore no SLNB (primary ALND) (n = 14)

• – Diagnostic excision biopsy or excision of DCIS which proved cancer, therefore primary ALND (n = 12)

• – Occult breast cancer (n = 5)

• – LN found with other imaging modality than US (n = 5)

• – Inflammatory breast cancer (primary ALND) (n = 4)

• – Other cancer in same breast/axil (n = 4)

• – Clinical metastasis to axil, nothing noted about US (n = 4)

• – Breast cancer found in relation to breast reducing surgery (n = 2)

• – No SLNB because of breast implants (primary ALND) (n = 1)

• – 2 macrometastases found in lumpectomy, nothing noted about SLNB (n = 1)

• – Patient from Greenland, no SLNB (n = 1)

• – Normal axil at time of diagnosis, alternative treatment for 1 year, clinical metastases at time of operation (n = 1)

• – Patient refrains from SLNB (primary ALND) (n = 1)

• – Tumor excision in other department showed breast cancer, hereafter unsuccessful fine needle aspiration x 2, primary ALND (n = 1)

• – Patient diagnosed in 2007, surgery in 2012 for unknown reasons (n = 1)