Axillary node dissection has long been a mainstay in the treatment of breast cancer: It provides precise staging and prognostication, prevents local recurrence in the axilla, and, in patients with positive nodes, may modestly enhance survivalall important goals in a disease that responds to both local and systemic therapy and has a long natural history. To date, no procedure has proven as effective as axillary lymph node dissection in accomplishing these goals. Nevertheless, axillary dissection is a major operation, requires general anesthesia, and produces long-term morbidity in a small, but significant, minority of patients.
Within the last 2 years, sentinel lymph node (SLN) biopsy (Figure 1) has rapidly emerged as the most exciting development in the surgical treatment of invasive breast cancer since the advent of breast conservation. It has the potential to identify those patients most likely to be helped by axillary dissection (ie, those with positive nodes) and to spare node-negative patients, who cannot benefit, from the morbidity of an operation. Undoubtedly, sentinel lymph node biopsy will rapidly become a standard treatment option for all patients with early-stage breast cancer and will replace axillary dissection for many of these patients.
After a brief description of the history of SLN biopsy, this review will discuss the feasibility and accuracy of the procedure, as well as some of the technical and clinical issues raised by this new technology. It will also attempt to answer a key question: Where will SLN biopsy ultimately fit into the treatment algorithm for breast cancer in the 21st century?
Most physicians are familiar with Sappeys classic studies of the lymphatic anatomy of the chest wall, based on mercurial injection of cadaver specimens. The lymphatics of the chest wall and breast converge into a few main trunks, which, in turn, drain into relatively few nodes low in the axilla. These studies formed the anatomic basis of the complete axillary dissection incorporated into radical mastectomy by Halsted in the 1890s. The internal mammary nodes represent an alternative route of lymphatic drainage, but, as shown in the work of Turner-Warwick and others, receive only a small fraction of the lymphatic flow of the breast.
The underlying simplicity of the lymphatic anatomy of the breast was emphasized in subsequent studies using direct lymphangiography.[5,6] As pointed out by Borgstein and Meijer in a recent comprehensive overview of the subject, work by Kett et al in 1970 identified a lymph node (which they called the Sorgius node) that received the initial drainage of contrast medium from the breast.
The phrase sentinel node must be credited to Cabanas, who, in 1977 described cannulation of the dorsal lymphatics of the penis as a means of identifying the first lymph node (sentinel node) draining penile carcinoma. He noted that the sentinel node was frequently the only positive node, and proposed that if the sentinel node were negative, a deep inguinal node dissection might be avoided.
In the 1980s, Morton and colleagues independently developed the sentinel node concept as an outgrowth of their work mapping the drainage patterns of cutaneous melanoma with lymphoscintigraphy. Their classic 1992 study described the intradermal injection of blue dye into the melanoma site and the identification of a blue node in the regional nodal basin. Among 237 node basins studied (with a standard node dissection performed in all cases), SLNs were identified in 82% of patients. Sentinel lymph node biopsy correctly predicted regional node status in 99% of successful procedures and 95% of node-positive cases (38/40).
As recently reviewed by Brady and Coit, 10 groups have since reported comparable or better results using SLN biopsy (blue dye and/or radioisotope localization) in melanoma.
The use of SLN biopsy in breast cancer was first reported in 1993 by Krag et al, who employed radiolocalization, and in 1994 by Giuliano et al, who employed blue dye. In the study by Krag et al, isotope identified the SLN in 82% of 22 cases, with 100% accuracy. In the report of Giuliano et al, blue dye identified the SLN in 65% of 173 cases, with 96% accuracy.
Since these pioneering reports, 14 groups have published their results with SLN biopsy for breast cancer, validated in all cases by a concurrent axillary node dissection. Of the 14 groups, 7 used isotope localization,[13-19] 3 used blue dye,[20-22] and 4 used a combination of both methods.[23-26] Table 1 summarizes the results of all 1,564 reported cases.
Success in Identifying the SLN
Regardless of method, SLNs were identified by all of the investigators in a large majority of cases. The 66% success rate for blue dye in the initial 1994 report of Giuliano et al may simply reflect the developmental stage of a new procedure; in this groups more recent experience, blue dye successfully identified the SLN in 93% of cases. Overall, radioisotope localization appears to find the SLN more frequently than does blue dye, and the combination of isotope plus dye appears to be superior to isotope localization alone in these pilot studies.
Accuracy in Predicting Axillary Node Status
Sentinel lymph node biopsy yielded an incorrect result in 2% of all patients (for an accuracy of 98%) and 5% of node-positive patients (for a sensitivity of 95%). Although nearly half of the series (7 of 16) reported an accuracy of 100%, these comprise only 23% of the total number of cases. No diagnostic test is perfect, and the accuracy of SLN biopsy is probably slightly less than 100%.
Validation of the SLN Hypothesis
As shown in Table 1, the SLN was the only site of nodal metastasis in 45% (range, 33% to 67%) of all node-positive cases, strongly supporting the SLN concept. If the SLNs were examined for micrometastatic disease with enhanced pathology (serial sectioning, with both hematoxylin and eosin [H&E] and immunohistochemical [IHC] staining), one might question whether equally close examination of the nonsentinel axillary nodes would also find micrometastases, undermining the uniqueness of the SLN as the true first node. Turner et al have conducted an elegant histopathologic validation of the SLN hypothesis (Table 2). Among 60 patients whose SLNs were negative on both frozen and serial sections (using H&E and IHC), all of whom underwent axillary dissection with serial sectioning of all of the remaining axillary nodes, only 1 (0.1%) of 1,087 nonsentinel nodes contained tumor.
Because the SLN technology is evolving rapidly, variation in technique is widespread, and anecdote rather than controlled observation has been the rule. Nevertheless, all of those who perform SLN biopsy would agree that it is a multidisciplinary procedure requiring close collaboration among nuclear medicine (for institutions using radiolocalization), surgery, and pathology. A summary of some pertinent technical issues relevant to each specialty follows.
Localization of the SLN represents a new challenge for the specialty of nuclear medicine, with requirements quite different from those of solid organ imaging. The behavior of injected radiocolloids is largely a function of particle size and interstitial pressure. The largest particles (> 200 nm) simply remain at the injection site, and the smallest (< 5 nm) rapidly disperse into the bloodstream. Particles between 5 to 10 nm in size rapidly enter the lymphatics but spread into numerous nodes. Sentinel lymph node localization depends on a small fraction of the injected isotope dose (perhaps 1%) migrating consistently to relatively few regional nodes, and optimal particle size is probably between 10 and 200 nm.
Interstitial pressure, which must be elevated for lymphatic uptake of particles to occur, is related to both route (intradermal vs intramammary) and volume of injection: Lymphatic uptake of isotope may be greater with either a high-pressure intradermal injection or a high-volume intramammary injection. American investigators have used technetium-99m sulfur(Drug information on sulfur) colloid,[11,18,23,25] while European studies have employed technetium-99m colloidal albumin.[13-17,24,26] The particle size in technetium-99m sulfur colloid preparations is particularly subject to wide variation; this may be reflected in the somewhat less consistent results reported for radiolocalization in the American literature.
One might assume that intramammary injection of isotope, as was done in the studies from America and Holland,[11,14-18,23-25] would most accurately emulate the lymphatic drainage of a breast cancer. Interestingly, in the two series from Milan,[13,16] lymphatic mapping based on a subdermal injection of isotope over the tumor site was equally successful and accurate in predicting axillary node status. The dermal and parenchymal lymphatics of the breast may simply drain to the same SLN.
Investigators using technetium-99m sulfur colloid typically inject 0.3 to 1.0 mCi in a volume of 4 mL of normal saline around the tumor (or biopsy site) 1 to 4 hours prior to operation,[11,18,23,25,26] whereas those using technetium-99m colloidal albumin inject 0.1 to 1.6 mCi either subdermally or into the breast about 24 hours in advance of surgery.[13-17,24] Filtration of technetium-99m sulfur colloid prior to injection produces greater uniformity of particle size, and has become standard practice in lymphatic mapping for melanoma. The smaller particles rapidly traverse the lymphatic vessels and regional nodes, helping to identify anomalous patterns of lymphatic drainage on preoperative lymphoscintigraphy.
In our experience performing SLN radiolocalization for breast cancer at Memorial Sloan-Kettering Cancer Center, we have noted a higher failure rate with filtered than with unfiltered technetium-99m sulfur colloid. Therefore, we continue to use an unfiltered preparation, as initially recommended by Krag et al.[11,18] In contrast, Albertini et al have reported good results with filtered technetium-99m sulfur colloid.
Lymphoscintigraphy is a well-established component of lymphatic mapping for melanoma. It identifies anomalous patterns of lymphatic drainage, which, in turn, directly alter the surgical approach. This examination is probably less useful on a routine basis in breast cancer, where the primary focus is the axilla. Management of the 14% of patients who demonstrate internal mammary drainage is a new, unresolved problem.
The surgeons objective is simply to find the SLN as consistently as possible. Each failed localization will result in an axillary dissection that might not otherwise have been necessary. Debate centers on which method is best to accomplish this objective: isotope, blue dye, or both? Table 1 suggests that all three approaches work.
With isotope, as first described by Krag et al, the surgeon uses a handheld gamma probe intraoperatively to find the axillary hot spot(s) corresponding to the SLN(s) and removes hot node(s) until the axillary background radiation count falls below a defined threshold level . With blue dye, as described by Giuliano et al, the surgeon identifies blue lymphatic vessels exiting the tail of the breast and traces them to a blue-stained SLN in the axilla, removing all blue nodes. All blue and/or hot nodes are removed and submitted for pathologic examination.
Early in the surgeons experience, SLN biopsy is best validated by the performance of a backup axillary dissection (as part of a formalized protocol) after removal of the SLN. The learning curve for this procedure varies from institution to institutions and surgeon to surgeon. At present, it is impossible to specify exactly how many procedures should be done with validation before SLN biopsy can be performed as a stand-alone procedure.
In our experience performing more than 800 SLN biopsies at Memorial Sloan-Kettering, we have found that failed SLN localizations diminish but do not altogether disappear with experience; the results of other investigators (Table 1) support this impression. We advocate the combined use of isotope and blue dye for SLN biopsy (Figure 2): Although 80% of SLNs were found by both isotope and dye, 10% were found by isotope alone, and 10% by blue dye alone. This additive effect of isotope and blue dye was first noted by Albertini et al, and was confirmed by our initial experience and that of Barnwell et al.
The surgeons greatest concern in undertaking SLN biopsy is that the SLN will prove to be falsely negative. False-negative SLN biopsies (like failed SLN localizations) also diminish over time. Our experience suggests that about half of all falsely negative SLN biopsies will occur within the first 10 cases performed by each surgeon.
No studies have specifically addressed the relative value of isotope and blue dye in finding the positive SLN. However, our experience suggests that while most positive SLNs will be found by both isotope and blue dye, a small fraction (perhaps 10%) will be found by either isotope or blue dye alone. Although others have achieved excellent results with a single modality (Table 1), the reliability of SLN biopsy in our hands would have been undermined by reliance on a single localization technique.
The most crucial element in SLN biopsy is enhanced pathologic analysis of the SLN. Reporting on the early experience with melanoma, Gershenwald et al found that 4.1% (10/243) of patients with negative SLNs (by routine single-section pathologic examination) later developed a local relapse in the undissected nodal basin. Reanalysis of the negative SLNs in these 10 patients (with serial sections and IHC stains) demonstrated that 80% were, in fact, positive.
Thus, an increasing number of investigators performing SLN biopsy for breast cancer have relied on serial sectioning of the SLN, with both H&E and IHC staining of each section. The IHC technique uses antibodies to cytokeratin and, thus, identifies epithelial cells that are presumed to represent metastases from the breast (Figure 3). Although no authors have reported the occurrence of a falsely positive SLN in breast cancer, benign rests of epithelial cells (typically melanocytes) in the subcapsular area of a lymph node may occasionally be mistaken for metastasis; we have encountered 1 such case in our first 600 SLN biopsies.
Reverse transcriptasepolymerase chain reaction (RT-PCR) technology has the exciting potential to identify metastases even smaller than those found by IHC, but thus far has proven to be problematic in the study of SLNs from breast cancer patients. First, since breast cancer has no unique marker (such as tyrosinase for melanoma), the analysis must depend on nonspecific epithelial cell products. As a result, there is no way to be certain that the amplified gene product obtained was expressed specifically by a tumor cell within the node.
In addition, the level of expression at which a result is defined as positive or negative is somewhat arbitrary and allows considerable latitude in the interpretation of results. Thus, RT-PCR for SLN analysis in breast cancer remains investigational at present.