According to previous ARM studies using dyes and radioisotopes, the incidence of ARM nodes detect in patient populations were reported to be 50% to 91% [8–14]. The detection of ARM lymphatic ducts is more important than that of ARM nodes because the aim of ARM is to spare lymphatic pathways, which is not the case for the SLNB procedure. Noguchi et al. reported that the identification rate of ARM nodes and/or lymphatics was 88% using fluorescence navigation with ICG, and concluded that the fluorescence imaging technique was useful for detecting not only the ARM lymph nodes, but also the lymphatic ducts . In the present study, the identification rates of ARM nodes and lymphatic ducts were 88% and 76% for patients without SLNB and 74% and 86% for patients with SLNB, respectively, which was similar to the results in previous studies. The fundamental concept of ARM procedure is based on the assumption that the lymphatic pathway from the upper extremity would not involve metastasis from the primary breast cancer [8, 11, 12]. However, the metastatic rates to ARM nodes were reported to be around 14% to 43% in recent feasibility studies [10, 13, 14]. The oncologic safety of preserving ARM nodes or lymphatic ducts is the most important issue for performing ARM in practice. Nos et al.  explained the assumptions concerning ARM node metastasis. The first assumption was that ARM nodes would belong to the central nodal group, which is highly related to the lymphatic pathway originating from the breast. If this was the case, the ARM procedure would have limited surgical benefit.
A further assumption was that the natural progression of breast cancer would lead a metastasis to ARM node following the other extensive nodal metastasis. In the present study, the patients with clinically node-positive axilla had a higher risk (24%) of having positive ARM nodes than those with clinically node-negative axilla (3%). Of 35 patients with SLNB, only one patient had a positive ARM node (Table 3). She had a silicone implant in the bilateral breast and it was difficult to estimate the nodal status using ultrasonography (US) and computed tomography before surgery. She had severe nodal metastasis (nine positive nodes) at levels I to III of the axilla.
In the seven patients with positive ARM nodes, four (57%) had extensive nodal metastasis (> 4 metastatic nodes). This result may support the notion that patients with breast cancer and extensive nodal metastasis would have positive ARM nodes following tumor progression. However, the other three patients also had a few positive nodes in the axilla. In these patients, there must be another reason why ARM node status was positive. In previous ARM studies, ARM nodes and/or ducts were observed in the sentinel biopsy field in 38% to 75% of all cases [11–13] and concordance between sentinel nodes and ARM nodes was reported in approximately 20% of the patients . The positive ARM node in noted in case 2 (Table 3) was not located in the sentinel lymph node biopsy field, which seems to equate to field B or C (Figure 2), but rather to field A, where ARM nodes were often observed (Figure 2). This case may suggest that some ARM nodes belong to the breast central nodes and that metastasis to ARM nodes could occur in the patients with a few positive axillary nodes. Therefore, it is important to confirm the metastatic status of ARM nodes in order to for those nodes to be retained during operation. Han et al. reported the use of FNAC or frozen biopsy of partial resection of suspicious ARM nodes to confirm oncologic safety if the surgeon suspected that those nodes were metastatic. Unfortunately, the detail information from that study concerning the FNAC results for ARM nodes is not available .
In practice, FNAC of axillary lymph nodes with US guidance is widely accepted as a useful procedure for providing preoperative information on nodal status . A recent systematic review of the US-guided FNAC shows a sensitivity that varies between 30.6% and 62.9% and a specificity of 100% . However, inadequate sampling is a potential limitation of FNAC for diagnosis. Ciatto et al. reported that a sampling error might be corrected by repeated sampling and showed that sampling was insufficient in 10.8% of cases, which is comparable or lower than the inadequacy rate reported in other studies. In our study, the sampling error was 26% in the without SLNB group and 18% in with SLNB group. Micrometastasis in a lymph node was one of the causes for false-negative result in a previous study of FNAC for axillary lymph nodes . However, the chief cause of our sampling error may be technical in nature because all patients had macrometastasis in the ARM nodes in the present study. Our sampling error rate was higher than that of FNAC by US-guided aspiration. Compared with an US-guided procedure, it may be difficult to obtain sufficient materials from lymph nodes in a surgical field using our method, because US imaging can be helpful in confirming the fine needle tract to the target. Repeat aspiration for ARM nodes may improve the success rate of sampling if sufficient material for diagnosis is not harvested after the first aspiration. The limitation of our study included the small number of enrolled patients and relative inexperience with the new procedure. Studies including a larger series of patients are required to determine the efficacy of FNAC for the preservation of ARM nodes during ALND.