Background and objective

The current goal in oncologic breast surgery is to tailor treatment options to allow optimal care without unnecessary interventions. Breast surgery has steadily evolved from an extensive to a more conservative approach. Since 2000, many clinical trials have confirmed that sentinel lymph node biopsy (SLNB) is an accurate technique that permits omitting a complete axillary lymph node dissection (ALND) in selected patients while diminishing the incidence of arm and shoulder morbidity [1, 2]. SLNB is developing in new directions (multifocal/multicentric tumors) [3]: applications with neoadjuvant chemotherapy [4–10], axillary reverse node mapping [11–13], and nanotechnology [14]. Axillary lymph node status is the single most significant predictive factor for patients with invasive breast tumors [15]. A positive sentinel lymph node (SLN) requires subsequent ALND. Sparing a second axillary surgery is a current concern that has stimulated the development of other approaches including molecular biology techniques for intraoperative assessment of SLN and preoperative detection of node metastases through the use of imaging methods. Ultrasound (US) is a simple and well-accepted method to examine axillary and non-axillary lymph nodes associated with breast cancer. Recent publications have reported that routine axillary US combined with cytology or core biopsy is an effective method to evaluate lymph node metastases prior to surgery [16–35]. Thus, pretreatment axillary US for early breast cancer patients and needle sampling of morphologically abnormal lymph nodes are now widely recommended [36]. The aim of our study was to establish the accuracy of US-guided fine needle aspiration cytology (FNAC) of axillary lymph nodes for the detection of clinically silent metastases and to find out how often a SLNB could be avoided.

Material and method

From January 2006 to December 2010, 144 consecutive patients with invasive breast cancer, clinical stage T1/2N0, were evaluated by axillary US with FNAC. Thirty-six patients were excluded due to a personal history of breast cancer, previous or ongoing chemotherapy or previous breast/axilla surgery, and non identifiable lymph node on axillary US. All 108 remaining patients underwent initial breast biopsy confirming invasive malignancy.

Both axillary US and node FNAC were performed by the same breast radiologist. A high-resolution probe (12 MHz electronically focused linear array transducer) of ATL HDI 500 Philips Healthcare was utilized. The suspicious ultrasound node characteristics in our study were: length/width ratio of <1.5, cortical asymmetrical thickness more than 3 mm, hypoechoic cortical nodule deforming hilum, and the absence of the fatty hilum. If at least one of these criteria was observed, the node was selected for FNAC. If more than one abnormal lymph node was found, the most suspicious one was selected. If all suspect lymph nodes were similar, the lowest one in the axilla was selected. If no suspect lymph node was detected, the lowest normal lymph node larger than 5 mm was sampled. In one case a preoperative breast NMRI identified a suspicious internal mammary lymph node that was sampled by US-guided FNAC.

All cytological samples were processed with the Thin Prep System and analyzed by a breast specialist pathologist and classified as: insufficient for diagnosis and negative or positive for malignancy. Patients with cytology that was negative or insufficient for diagnosis were referred to SLNB. Patients with positive cytologies and all patients treated with neoadjuvant chemotherapy underwent ALND. If the intraoperative imprint cytologies (Diff-Quick staining protocol) or the extemporaneous frozen sections of SLN (hematoxylin and eosine staining method) were positive for malignancy the patients underwent immediate ALND. If the SLNB were negative based on extemporaneous examination but micro or macro metastatic on final histology (metastasis, ≥0.2 mm) the patients underwent ALND within 2 weeks. If isolated tumor cells were identified through definitive histology, no further ALND was performed.

The total number of harvested nodes, the number of positive nodes, and the size of metastasis were recorded. The tumor size and grade, histological type, lymphovascular or perinervous infiltration, the type of surgical intervention, and the neoadjuvant treatments were also included in our data. The final pathological results of harvested sentinel lymph nodes or ALND were correlated with US-guided FNAC. The sensitivity, specificity, and positive and negative predictive values were calculated.

Findings

One hundred eight patients had axillary lymph US-guided FNAC as a part of the investigation for invasive breast cancer. No immediate or late complications such as bleeding, hematoma, nerve injury, or infection were reported. Eighty-six patients underwent primary surgery and the other 22 underwent neoadjuvant chemotherapy with subsequent surgery. The median age was 58 (range 33–83) and 54 (range 33–71) years for the two groups, respectively.

The most frequent histological type of primary invasive breast tumor in our group was ductal carcinoma in 82 (76%) patients. Other described histological types were lobular carcinoma in 15 (13.9%) patients, mixed carcinoma in 9 (8.3%), and mucinous carcinoma in 2 (1.8%) cases.

As overlap of ultrasound features between reactive nodes and suspicious/metastatic ones is documented [19, 37], our breast radiologist decided to puncture also non-suspicious nodes larger than 5 mm [19, 38]. As our study is retrospective, the ultrasound reports were nonuniform and did not mention in detail the sonographic nodes description and the reasons for choosing a suspicious or a non-suspicious node. Hence, statistical analysis including the significance of each sonographic suspicious finding could not be made.

For 55 (50.9%) patients the FNAC was positive and for 46 (42.6%) patients it was negative. In seven (6.5%) cases the FNAC was insufficient for diagnosis (in one case the FNAC sampled only necrotic cells and in six other cases there was insufficient material to reach a conclusive result). All these patients were considered to be negative for statistical analysis. False-negative results were documented in 18 (33.9%) out of these 53 patients (46 with negative cytology and 7 considered negative, but insufficient for diagnosis). Three patients had isolated tumor cells on definitive histology despite a negative FNAC. We did not consider these results as cytohistological discordances as pathological classification remains pN0(sn) and no further ALND was performed.

For the patients in the primary surgery group, no false-positive result was documented. In the neoadjuvant chemotherapy group, 6 out of 18 patients with positive pretreatment FNAC had negative axillary lymph nodes. We have interpreted this apparent discrepancy as a complete node histological response and not as false-positive results. Figure 1 shows the patients' recruitment from 2006 to 2010. Figure 2 describes our study design and the results. Table 1 reviews the final histological findings for the 86 patients treated by primary surgery. Table 2 compares the surgical node staging with FNAC for the primary surgery group. Table 3 summarizes the sonographic aspect of the punctured nodes and the cytological results in the same group. The data in Tables 4 and 5 review the clinicopathological features of invasive carcinomas and the 14 cytohistological discordances in primary surgery group. For this group of patients, the cytohistological concordance was established in 72 (83.7%) patients (37 with positive and 35 with negative results). Consequently, for 37 patients a second axillary surgery was avoided (43.02%). Fourteen discordances were documented as either FNAC negative or insufficient for diagnosis but positive on definitive histology. In 12 patients, the FNAC was negative and in two patients the material was insufficient for cytological diagnosis. Among these 14 patients with cytohistological discordances, 11 had only one positive node, 2 patients had two positive nodes and 1 patient had three positive lymph nodes.

Fig. 1
figure 1

Annual patient recruitment between 2006 and 2010

Full size image
Fig. 2
figure 2

Study design and results

Full size image
Table 1 Surgical node staging compared with preoperative FNAC (108 patients)
Full size table
Table 2 Surgical node staging compared with preoperative FNAC in primary surgery group
Full size table
Table 3 Lymph node morphology by axillary ultrasound and cytological findings in the “primary surgery” group
Full size table
Table 4 Clinicopathological features of invasive carcinoma primary surgery group
Full size table
Table 5 Cytohistological discordances in primary surgery group
Full size table

The internal mammary lymph node sampled by the US-guided FNAC was positive for malignancy. The patient underwent ALND (14 negative axillary lymph nodes) and the final pathological stage was pT1cN2b.

Discussion

The appropriate pretreatment evaluation enables personalized management of breast cancer patients. SLNB is widely accepted today and has led to decreased shoulder and arm morbidity. Nevertheless, this technique requires the sustained coordination of a multidisciplinary team. Some patients need a second axillary surgery if the definitive SLNB with immunostaining proves the presence of node metastases. Although there are still controversies concerning morbidity differences between immediate and delayed ALND [39–43], operating on a distorted axilla represents a surgical challenge with emotional distress for the patient as well as considerable additional medical costs [44]. Axillary ultrasound is noninvasive, reproducible, largely available, widely accepted by patients, and cost effective [35, 44]. In addition, all lymph node chains can be evaluated (intramammary, internal mammary chain, infra- and supraclavicular, or axillary nodes).

In our study, immediate assessment of the quality of FNAC was not feasible. Seven (6.5%) out of 108 FNAC specimens were insufficient for diagnosis. Immediate assessment of specimens by a pathologist might reduce the proportion of inadequate samples [45] and allow additional lymph node passes by the radiologist to improve quality [28]. For the 86 patients treated by primary surgery, the overall sensitivity was 72.55% and the specificity was 100% (Table 6) although identifiable non-suspicious nodes were punctured and aspirated for cytological pretreatment examination. Forty-six patients of this group had sonographic suspicious nodes. Thirty-one out of these 46 nodes were positive. By puncturing normal nodes as well, a total of 37 patients with node metastasis were identified (Table 3). The major benefit of US-guided FNAC was that it spared a second surgical intervention for these 37 patients (43% of the primary surgery group). For patients with neoadjuvant treatment, this approach enables the gathering of further information about the in vivo response to chemotherapy. Patients who have a complete nodal and breast pathological response enjoy a much better outcome than those who still have residual disease after therapy [46].

Table 6 Sensitivity, specificity, positive predictive value and negative predictive value
Full size table

Recently, the NCCN guidelines (version 2.2011) introduced prechemotherapy SLNB as an option for clinically node-negative breast cancer patients benefiting from neoadjuvant regimens. As for early breast cancer patients, we expect that FNAC will allow a better selection of patients for prechemotherapy SLNB and will carry the same prognostic value as SLNB if complete histological response is observed.

False-positive results have seldom been reported [29, 47]. These results have been related to misinterpreting the presence of reactive lymphoid or mesothelial cells as carcinoma infiltration. The majority of cases believed to be false-positive may, in fact, be true-positive (complete histological response or failure to detect minimal residual disease in the final histology). We believe that all positive FNAC with a negative definitive histology must be treated cautiously in order to exclude a non-harvested positive node.

False-negative FNAC results probably occurred in part due to failure to target the real SLN or the most suspected region of the lymph node and perhaps to misinterpretations including failure to recognize tumor cells. As this is a retrospective study, no information about the punctured lymph node and the correlation with the SLN surgically removed was possible.

Conclusions

With a sensitivity of 73%, axillary US associated with FNAC plays an important role in preoperative lymph node staging for newly diagnosed invasive breast cancer patient. In selecting non-suspicious nodes for pretreatment lymph node cytology, we could spare a second axillary surgery for 43% of the patients. In addition, for patients included on neoadjuvant regimens, it allows a better selection for prechemotherapy SLNB and furthermore an evaluation of the histological response to treatment offering valuable information on patient prognosis and predictive factors.