Breast cancer is the most common female cancer in Belgium and many other countries [1]. In 2014, 10,557 women in Belgium were diagnosed with breast cancer. One of the most important prognostic factors is the status of the axillar nodes at the moment of diagnosis. The least invasive way to detect involvement of the axillary lymph nodes is the sentinel lymph node biopsy (SLNB) [2-4]. This staging technique is standard for women in an early stage of mammary carcinoma and replaces the axillary lymph node dissection (ALND) if the sentinel lymph node (SLN) is negative [5,6]. The SLNB is an attractive alternative for the ALND as the latter involves considerable morbidity [7]. This morbidity consists of temporary as well as lifelong complications: pain, swelling, lymphedema, stiffness, weakness, numbness, limited shoulder mobility, nerve damage, and infections [8-10]. Furthermore, the SLNB has less morbidity, is cheaper, and less time-consuming than the ALND and it has a similar long-term outcome [6,10,11]. Contraindications for SLNB are patients with a clinically positive axilla (N1), with a locally advanced tumour, and axillary metastases who have undergone induction chemotherapy and pregnant women [5]. A local recurrence or prior axillary procedure is regarded an indication and no contraindication [4,12]. Lately more and more researchers questioned whether the SLNB could also be used in women with locally advanced mamma carcinoma. Presently most of such patients receive initial neo-adjuvant chemotherapy (NACT) before surgery [13]. The criteria [14] for NACT administration are: tumour size > 2 cm or clinical involvement of the axilla. NACT may also be beneficial for mammary carcinoma in an earlier stage [15].

The advantages of NACT compared to postoperative chemotherapy are: 1) reduction in the size of the original tumour (±80% of the patients) thereby increasing the possibility of breast conserving therapy [16], 2) conversion of axillary lymph nodes to a negative status (20-40% of the affected lymph nodes) [16], 3) determination of the in vivo chemo-sensitivity of the tumour [17], and 4) longer disease-free survival after a complete pathological response of the primary tumour after NACT [13,14,18-23]

To determine the status of the axilla and to ensure that every involved axillary lymph node is removed, patients normally undergo ALND after NACT. As described earlier, ALND has considerable postoperative complications, which makes one wonder whether SLNB can also be used after NACT. In cases where SLNB is negative, ALND can be avoided. This article reviews clinical studies studying the feasibility and accuracy of the SLNB after NACT. By integrating several studies, a greater patient population is created, resulting in a more accurate estimation of the identification and false negative rates of SLNB after NACT.

The electronic databases used for the search were PubMed, Medline, and the Cochrane Library. ‘BREAST CANCER', ‘NEOADJUVANT CHEMOTHERAPY', ‘SENTINEL LYMPH NODE BIOPSY', ‘AXILLARY CLEARANCE', and ‘MORBIDITY' were used as search terms (MeSH terms). The search extended from 1994 to 2016 providing a study period of 22 years. 1994 was chosen, as this was the year in which the SLNB was introduced. Since then, the procedure has gained popularity and has been considered for implementation in the diagnosis and treatment of locally advanced breast cancer. The parameters of interest from relevant studies were pooled. For SLNB to be acceptable the criteria of a false negative rate of less than 5% and an identification rate of more than 90% were set [15].

Articles were selected by inclusion criteria listed in Table 1. Two independent readers reviewed found literature. Relevant information was extracted from each individual study and compounded in a structured table. Parameters of use were identification rate, false negative rate, sensitivity, negative predictive value, and accuracy. Pooled identification and false negative rate were calculated to assess whether the criteria [15] were met.

In Figure 1 a flow chart of the search method is shown. Table 2 gives and overview of the selected studies and Table 3 shows studies on SLNB after NACT with N1 at presentation.

Figure 1.

— Flowchart of the search method.

The most recent guidelines [15] dictate SLNB to be a feasible method in predicting the axillary status if an identific ation rate (IR) ≥ 90% is achieved. The pooled IR was 88%( Table 4), which did not meet the criterion of an IR ≥ 90%. Even if studies with less than ten patients per year, which would suggest less experience with the SLNB technique, were excluded and the IR did not change significantly (88%). Looking at only studies which included patients with documented N1 disease before NACT, an IR of 88% could be maintained.

The pooled false negative rate (FNR) obtained from all 43 studies was 12%. Difficult excluding studies with less than ten patients a year did not alter this result. As for studies only including patients with N1 disease at presentation, the pooled FNR remained at 12%. According to the recent guidelines [15], a SLNB was feasible in predicting the axillary status if the FNR ≤ 5% -a requirement which none of the Tables (2 and 3) met.

Controversy to current beliefs of no difference in IR or FNR was seen in extracting studies looking only at patients with axillary involvement at presentation. This might reflect the possibility that documented N1 disease before NACT does not affect IR or FNR.

Knowledge of the accuracy and feasibility of the SLNB after NACT as studied in this review which gives weight to the various arguments in favor and against this procedure (Tables 5 and 6).

This comparison showed that implementing the SLNB after NACT has an important advantage compared to SLNB before NACT. The variables that could influence the feasibility and accuracy of the SLNB after NACT are discussed below.

The first variable is the axillary nodal status at presentation. Different studies [40,44,47] documented a significant fall in the IR in patients with a clinical positive axilla at presentation. Gimbuerges et al. [41] showed a FNR of 30% in patients with a clinical positive initial status compared to 0% in a negative status. This led to the conclusion that the initial axillary status has an important influence on the FNR after NACT. This was also confirmed in the SENTINA study [48] where the overall FNR was 14.2% for patients whose axillary status converted. Yet, other studies [17,38,44] reported no significant effect of the initial axillary status on the FNR. Results in this review support this theory, as there was no difference in pooled IR's or FNR's if studies were selected with patients with N1 disease at presentation.

The primary tumour size (PTS) is cause of significant controversy. Several studies [38,40,47,49] showed a significant lower IR for larger primary tumours. Similarly two studies [32,41] showed a higher FNR in larger primary tumours. Mamounas et al. [17] showed a similar phenomenon: for a PTS < 2 cm the FNR was 5.0%, for a PTS between 2-4 cm 9.7%, and for a PTS > 4 cm 13.8%. However these differences did not reach statistical significance. These results showed that the PTS had a negative influence on the feasibility of SLNB after NACT. Another study [16] explained this by documenting that larger primary tumours had an increased risk of infiltration of the lymph vessels by malignant cells and thus enhancing a failed identification.

One study [49] showed a significant fall of the IR with increasing age. The average age in patients with a successful IR was 44.9 and 53.5 (p = 0.007) for patients with a failed IR. Another study [47] showed average ages of 47.0 and 51.3 years (p = 0.004), respectively. This could be explained by replacement of the lymph nodes by fat as people age. The same could be said with increasing BMI where accumulation of fat in the lymph vessels prevents the uptake of the radioactive tracer. Hino et al. [40] showed that the IR for patients with a BMI >25 decreased significant with failed identification in 56%. Increasing the injection volume or breast massage after injection could solve this problem. Another study [32] however could not demonstrate a significant relation with increasing BMI.

Surgical experience was a very important factor as well, as shown by different studies [25,29]. SLNB failures were seen more often earlier on in the studies due to less surgical experience with the SNLB procedure itself. A surgeon is considered experienced when he has performed a sufficient number of procedures achieving an IR of ≥ 90% and a FNR of ≤ 5%. Schwartz et al. [31] followed a study design with the objective that a single surgeon performed every procedure so the IR depended on his experience alone. The same IR's were achieved for the same surgeon, showing that the NACT didn't influence the IR.

As mentioned earlier, NACT could cause histological changes resulting in down-staging of the axillary nodal status. One study [39] found a significant fall in sensitivity of the SLNB in patients with a complete response. This was questioned because of the low number of false negatives and was not considered reliable. However, a lower sensitivity was documented by another study [17] for patients with a partial tumour response. This was reinforced by Brown et al. [26]: patients, who had a positive axillary nodal status before NACT and became negative after NACT, without histological changes, had a higher risk of a false negative axillary nodal status. The FNR decreased from 22% to 4.1% if these patients were excluded. Therefore, it could be possible to distinguish false negatives from true negatives, based on the absence or presence of these histological changes.

The FNR appeared to be related to the number of SLN removed. Brown et al. [26] came to the following conclusion: if ≤ 3 SLNs were removed (n=56) a FNR of 33% was seen. However, if ≥ 4 SLNs were removed (n=30), an FNR of 5% was achieved. Other studies [23,50] also supported this finding and showed that there was a higher risk of a false negative axillary nodal status if < 2 SLNs were removed. This was also mentioned in the SENTINA study [48], where they proved a significant relation (p = 0.008) between the number of removed SLNs and the FNR. If one node was removed, an FNR of 24.3% was observed, contrary to the FNR of 6.1%, which was achieved if five SLNs were removed. However, if the number of SLNs removed was more than three, a FNR of less than 10% was achieved. Another study, ACOSOG Z1071 clinical trial [51] confirmed this. The study stated that FNR significantly decreased when at least three SLN's were examined (p = 0.07: 9.1% vs. 21.1%). In a recent study of Park et al. [52] including only node-positive patients, the overall FNR of was 7.8%. A median number of four SLNs was retrieved. Excluding patients with less than three removed SLNs showed a FNR of 6.1%. This marks the importance of the number of SLN removed, stating the theory of replacing the ALND with the SLNB if at least three nodes are identified.

The SN FNAC study [53] suggested a lower FNR could be achieved with mandatory use of immunohistochemistry for SLN evaluation, stating SLN metastases of any size, including micrometastases and isolated tumor cells should be considered positive. Following the theory, they obtained a low FNR (8.4%) by IHC evaluation because the rate of positive non-SLNs is higher and independent of the size of SLN metastases, contrary to patients who do not receive NACT, suggesting that axillary pathological response to NACT can be non-uniform, which means SLNs can already be converted, whereas secondary lymph nodes still contain residual disease.

The last critical point to be considered is whether ALND completion can be omitted in certain cases. The St. Gallen consensus conference [54] and the German-Austria-Swiss consensus conference [55] came to the same conclusions: 1) ALND can be omitted in the presence of isolated tumour cells (ITC) and micrometastases in the SLN. Confirmation by Galimberti et al. [56], found a five-year disease-free survival of 84.4% in the ALND group and 87.8% in the non-ALND group, stating that disease-free survival in the group without ALND was non-inferior to the ALND group (HR 0.78). Also, overall survival did not differ between the two groups (97.6% in the ALND group and 97.5% in the non-ALND group). 2) ALND can be omitted in patients with one or two SLN macrometastases, adequate systemic therapy, and low risk (pT1/2, estrogen receptor-positive, HER2/neu-negative, G1 or 2, postmenopausal). Even though the five-year recurrence rates in the ACOSOG Z0011 trial [57] were 0.5% and 0.9% for axillary dissection versus no axillary dissection, respectively, there was no significant difference in overall and disease-free survival. The question remains whether these omissions of ALND can be implemented after NACT.

Considering the previous facts, in combination with the results of Tables 2 and 3, it seems that the implementation of the SLNB after NACT is a worthy alternative for the ALND in patients with a negative axilla after NACT. The SLNB can, in this way, be performed at the same time of the lumpectomy or mastectomy, thus limiting surgery to one operation. If a patient is N1 at presentation (FNAC) but becomes N0 (SLNB) after NACT, ALND can be avoided if the SLNB procedure meets the criteria (IR > 90%; FNR < 5%). If a patient remains N1, ALND can only be avoided in certain cases [54,55]. The pooled IR of 88% and FNR of 12% (Table 3) calculated in this article, do not meet the criteria (IR ≥ 90%, FNR ≤ 5%) for SLNB after NACT to be accepted as a standard. Comparing these results with large studies concerning the feasibility and accuracy of the SLNB without NACT the following can be noted. These studies [58-60] achieved IR's of 96-97% and FNR of 5-7.3%. The IR (88%) after NACT did not differ much from the IR (96-97%) without NACT. The FNR (5-7.3%) however doubled and almost tripled after NACT (12%). However if certain guidelines are followed, such as retrieving three or more SN's and including immunohistochemistry staining, a more acceptable FNR could be reached. In these cases a SLNB could be considered as a as standard procedure after NACT. Studies with long-term follow-up are needed in order to determine the impact of this approach.

The authors thank Patrick Vandekerckhove, M.D. at St. Mary’s Hospital, Isle of Wight, for reviewing the English translation of the article.