It has long been known that lymph node status is one of the most powerful predictors of survival and prognosis for most malignancies.1, 2 Lymph node dissection and pathologic assessment are important components of the management of patients with primary cutaneous melanoma who have no clinically evident extracutaneous disease.3 The goals of lymphadenectomy for patients with melanoma are similar to those for patients with other cancers: staging, regional control, and, hopefully, increased survival. The optimal approach to the management of regional lymph nodes in patients with American Joint Committee on Cancer (AJCC) Stage I and II (localized) melanoma remains a controversial topic.4 One theory is that cutaneous melanoma commonly follows an orderly pattern of progression, with lymphatic dissemination to regional lymph nodes before the onset of distant metastasis.5, 6 This became the basis for performing elective regional lymph node dissection, the goal of which was to remove early metastatic disease and thus prevent distant metastasis.7 Results of several large studies disputed the therapeutic effectiveness of elective lymphadenectomy in improving patient survival.8, 9 Regional lymphadenectomy is associated with moderate morbidity; accordingly, it was subsequently recommended only for patients who have clinical evidence of metastatic melanoma in their regional lymph nodes.10 Although the ability of elective lymphadenectomy to enhance survival remains controversial, it clearly provides important staging information and enhances regional disease control. Optimally, elective lymphadenectomy should be performed only on patients who actually harbor occult lymph node metastases; this is where sentinel lymph node biopsy may be of value. Sentinel lymph node biopsy has little associated morbidity and is an effective means of identifying patients with clinically occult regional lymph node metastases.11, 12 Currently, patients with histologic evidence of metastasis in their sentinel lymph nodes are treated with selective regional lymphadenectomy followed by systemic adjuvant therapy.3, 13 Patients with histologically negative sentinel lymph nodes receive no further treatment.
The technique of sentinel lymph node biopsy was developed as a minimally invasive, low morbidity modality of lymph node status assessment for patients with primary cutaneous malignant melanoma.14 The sentinel lymph node is defined as the first lymph node or lymph nodes in a lymph node basin draining from the site of primary melanoma. Theoretically, in most cases it would be the first site of melanoma metastasis. This first or “sentinel” lymph node serves as a filter for metastatic melanoma cells between the primary cutaneous site and the remainder of the body. Previous studies have reported that the sentinel lymph node status is an accurate reflection of the lymph node basin status as a whole and that treatment decisions can be made based on the status of the sentinel lymph node alone.4, 15 In 223 patients who underwent sentinel lymph node biopsy followed by elective lymph node dissection, of patients with metastatic melanoma in their regional lymph nodes, 2.5% had histologically negative sentinel lymph nodes.14 As staging and therapeutic decisions are based on lymph node status, accurate pathologic assessment of the sentinel lymph node is of paramount importance. Pathologists routinely examine a lymph node specimen by bisecting the lymph node along its long axis, embedding the two halves of the lymph node on the cut surfaces, and examining one hematoxylin and eosin–stained section from each cut surface. Although data exist to indicate that this method underestimates the true incidence of metastasis,14, 16–18 the notion that routine methods underestimate metastasis may not be generally accepted. In addition to level sections through the tissue block, some observers recommend immunohistochemical stains for S-100 and HMB-45 in the analysis of SLN. Although S-100 and HMB-45 are the most commonly used markers in the detection of melanoma, they are inadequate because S-100 lacks specificity and HMB-45 lacks sensitivity. Antibodies against NK1C3 and, more recently, MART-1 have been added to the tools used by pathologists to detect melanoma. NK1C3 is more sensitive than HMB-45 but has the same specificity problems associated with S-100. MART-1 (melanoma antigen recognized by T cells) is a recently described, apparently specific, widely shared melanoma epitope localized to melanosomes and endoplasmic reticulum in benign nevi and melanomas.19 The most recent studies suggest that in some cases of metastatic melanoma, MART-1 is a more sensitive marker than either S-100 or HMB-45.20 Staging information and important therapeutic decisions are based on the presence or absence of metastatic melanoma in regional lymph nodes.
The purpose of this study was to determine whether a more detailed pathologic analysis of sentinel lymph nodes would increase the frequency with which metastatic melanoma is identified. In addition, we examined a panel of immunohistochemical reagents, including S-100, HMB-45, NK1C3, and MART-1, to determine which are the most helpful in the detection of microscopic melanoma metastases.
- Top of page
- MATERIALS AND METHODS
One hundred seven consecutive melanoma patients with reports of pathologically negative sentinel lymph nodes were identified. Of these, 12 did not have sufficient tissue for further study and were excluded from the study. One additional case was excluded from the study because of metastatic disease (in transit) at the time of sentinel lymph node biopsy. In the histopathologic analysis of sentinel lymph nodes, we initially evaluate the routinely processed hematoxylin and eosin–stained sections. In the event that metastasis is detected in transit, in lymph nodes, or otherwise, additional immunohistochemical studies are not performed. Thus, the protocol instituted in this study would not apply to a patient with in-transit metastasis identified at the time of sentinel lymph node biopsy.
Of the 94 patients admitted to the study, there were 56 men (60%) and 38 women (40%). The patient age at the time of primary melanoma diagnosis ranged from 19 to 81 years (mean age, 48 years; median age, 47 years).
A total of 280 paraffin embedded tissue blocks from 235 sentinel lymph nodes were studied. The number of blocks per case ranged from 1 to 19 (mean, 3; median, 3). The number of sentinel lymph nodes per case ranged from 1 to 9 (mean, 2.5; median, 2).
Most of the primary cutaneous melanomas occurred in the extremities (50 of 94 cases, Table 1) and half of the melanomas were of the superficial spreading histologic type (47 of 94 cases, Table 2). The vast majority of the melanomas were invasive to Clark level 4 (71 of 94 cases, Table 3). The mean Breslow thickness of the primary cutaneous tumours was 2.12 mm (range, 0.48– 12 mm; median, 1.7 mm; Table 4).
Table 1. Anatomic Distribution of the Primary Melanomas
|Site||No. of patients (%)|
|Upper extremity||31 (33)|
|Lower extremity||19 (20)|
|Head or neck||12 (12)|
Table 2. Histologic Subtypes of Melanoma of the 94 Patients in This Study
|Histologic subtypes||No. of patients (%)|
|Superficial spreading melanoma||47 (50)|
|Nodular melanoma||31 (33)|
|Lentigo maligna melanoma||3 (3)|
|Acral lentiginous melanoma||2 (2)|
|Desmoplastic/neurotropic melanoma||1 (1)|
Table 3. Clark Level Distribution of the 94 Primary Cutaneous Melanomas
|Clark levels||No. of patients (%)|
Table 4. Breslow Thickness of the 94 Primary Cutaneous Melanomas
|Breslow thickness||No. of patients (%)|
|<0.76 mm||5 (5)|
|0.76–1.50 mm||31 (33)|
|>1.50 mm||58 (62)|
Occult Metastatic Melanoma
Microscopic foci of metastatic melanoma were identified in 11 patients, resulting in a false-negative melanoma detection rate of 11.7% with the routine method of pathologic assessment. The mean primary tumor thickness of these patients was 2.2 mm (range, 1.04–4.5 mm; median, 1.7 mm). All of the metastatic melanomas were microscopic in size (smaller than 1.5 mm) and in 10 of 11 cases were located in the subcapsular sinus (Table 5). Five cases had focal involvement of the lymph node medulla. The melanoma deposits demonstrated nuclear pleomorphism, abundant cytoplasm, and fine melanin pigment. As a general rule, the foci of melanoma deposits were subtle and very difficult to detect on the hematoxylin and eosin–stained sections. All of the micrometastases stained positively with antibodies to S-100 protein, NK1C3, and MART-1 (Fig. 1). Three of 10 micrometastases did not stain with antibodies to HMB-45. There were also qualitative differences in the staining patterns of these various antibodies. Antibodies to S-100 protein stained dendritic reticulum cells, which were prominent in these lymph node samples. Small nerve twigs, adipocytes, and occasional histiocytes were also stained with antibodies to S-100 protein. There was little background staining of the lymph node parenchyma. Due to the small size of the melanoma deposits, these foci were often obscured by the strongly S-100 positive dendritic reticulum cells, making interpretation of the staining results more problematic (Fig. 1b). In a few cases, the S-100–stained sections were initially read as equivocal for melanoma. It was only after the detection of positively staining melanoma cells in the MART-1 sections that reexamination of the S-100 sections led to the identification of the S-100 positive melanoma cells. Antibodies to NK1C3 demonstrated a lack of specificity similar to that of S-100. Staining for NK1C3 stained mast cells, eosinophils, occasional histiocytes, and endothelial cells (Fig. 2). Staining for HMB-45, although highly specific, lacked sensitivity and was not present in over 30% of the micrometastases. Furthermore, the HMB-45 positive cases often only showed focal staining for HMB-45 in a few melanoma cells (Fig. 3). Staining for MART-1, on the other hand, was highly specific and sensitive. There was no background staining or staining of nonmelanocytic cells within the lymph node. We found MART-1 to be the most helpful antibody in the detection of microscopic melanoma metastases.
Table 5. Characteristics of Cases Found to Contain Microscopic Metastatic Melanoma
|Case no.||Immunohistochemical Profile||Location||Greatest Diameter of a Single Deposit||Breslow Thickness of Primary Melanoma|
Figure 1. Subcapsular deposits of metastatic melanoma in a sentinel lymph node are shown with the following staining. (a) H & E, (b) S-100, and (c) NK1C3. In addition to staining the subcapsular deposits of melanoma, both of these antibodies also stained scattered histiocytes.
Download figure to PowerPoint
Figure 3. HMB-45 staining of melanoma is shown. Some melanoma cells show an intensely strong brown signal for HMB-45; in other melanoma cells, the signal is weak or absent.
Download figure to PowerPoint
Capsular melanocytic nevi were detected in 9 lymph nodes from 8 patients (8.5% of patients). None of these 9 lymph nodes contained metastatic melanoma. All of these cases showed expansion of the lymph node capsule by sheets, nests, or single-cell arrays of bland, small, melanocytic nevus cells (Fig. 4). Nevomelanocytes were also occasionally observed extending along the fibrous trabeculae of the lymph node, but they were not observed in the lymph node parenchyma. All nevi showed staining with antibodies to S-100, NK1C3, and MART-1 (Fig. 4, Table 6). Stains with antibodies to HMB-45 were uniformly negative in subcapsular nevi.
Figure 4. Intracapsular melanocytic nevus is shown in a sentinel lymph node. (a) Benign nevomelanocytes expand the fibrous lymph node capsule in an H & E–stained section. (b) The capsular nevus cells stain positively for S-100, (c) NK1C3, and (d) MART-1.
Download figure to PowerPoint
Table 6. Immunohistochemical Profile of Capsular Melanocytic Nevi
|Case no.||Immunohistochemical profile|
In all cases of metastatic melanoma and capsular nevus, the positively staining cells were identified on the adjacent hematoxylin and eosin–stained sections.
Cases That Posed Interpretative Difficulties
There were two cases that posed diagnostic difficulties. One case showed a suspicious focus on immunohistochemical stains that was not present on the adjacent hematoxylin and eosin–stained sections. In this case, 2 cells staining strongly with the MART-1 antibody were present within the lymph node capsule. This was interpreted as benign (nevus), as the cytology of the cells appeared banal. The other case that was difficult to interpret showed approximately 10–20 cells in dishesive aggregates with abundant cytoplasm within the subcapsular sinus; staining was strong with antibodies to NK1C3 but not with antibodies to S-100, MART-1, or HMB-45. Examination of the adjacent hematoxylin and eosin–stained sections showed that these cells had the cytologic characteristics of macrophages or histiocytes.
- Top of page
- MATERIALS AND METHODS
Lymph nodes are traditionally examined for microscopic metastases by examination of one hematoxylin and eosin–stained section of the cut surfaces. In some settings, this method has been shown to underestimate the real incidence of metastasis.17, 21 In 1948, it was shown that by performing serial sections on axillary lymph node specimens from patients with breast carcinoma, an additional 33% of cases initially diagnosed as negative for metastases were found to be positive.17 With the advent of immunohistochemistry, it was found that the use of epithelial membrane antigen and keratin antibodies can increase the detection rate of breast carcinoma micrometastases in lymph nodes by 15%21 Nevertheless, in studies of breast carcinoma patients, it remains unclear whether the finding of microscopic metastases is associated with a worse prognosis.22–24 Data from a prospective randomized melanoma trial indicate that metastatic melanoma is detected histologically in approximately 20% of patients undergoing elective lymphadenectomy, whereas 22–37% of patients who were randomized to observation ultimately developed clinical evidence of regional lymph node metastasis.25, 26 More recently, the techniques of cell culture27 and polymerase chain reaction (PCR) were applied to the detection of microscopic melanoma metastases in lymph nodes.28–30 Of these various methods of detection, immunohistochemistry and additional hematoxylin and eosin–stained sections remain the most practical modalities for the detection of micrometastases. Cell culture methods take 4–6 weeks to give results and are highly dependent on the viability of tumor cells and the availability of an optimal cell culture environment. In addition, no permanent histologic record of the tumor cells in situ remains, because the whole lymph node is submitted for culture. Moreover, cell culture is an expensive and labor-intensive technique. Nevertheless, in a study of the lymph node culture technique, 31% of patients who were histologically lymph node negative were found to have melanoma cells in cultures derived from half of each excised lymph node. Although it is not a practical technique, these data support the concept that traditional techniques for analyzing lymph nodes underestimate the incidence of regional metastasis in patients with melanoma. RT-PCR of tyrosinase mRNA has identified a subset of patients whose lymph nodes are histologically negative yet contain molecular evidence suggesting the presence of metastatic melanoma. Unfortunately, RT-PCR examination for tyrosinase mRNA cannot differentiate between benign nevus cells and melanoma cells. This is theoretically a significant problem, given that we identified melanocytic nevi in 8.5% of the sentinel lymph nodes we studied. RT-PCR examination for other tumor-associated transcripts may allow differentiation between benign nevus cells and melanoma. Currently, histologic examination remains the only reliable means of differentiating between the nevus and melanoma. The architectural arrangement of the cells in question, the location of these cells, and whether cellular atypia is present cannot be determined by RT-PCR. These histologic attributes can only be adequately assessed by the examination of hematoxylin and eosin–stained sections. It is impossible to compare the sensitivity of cell cultures or PCR methods and histologic techniques, because one cannot histologically analyze multiple levels and simultaneously test the exact same tissue for melanoma cells using the cell culture techniques or PCR. To compound the problem, previous studies indicated that lymph nodes from patients with melanoma have a higher incidence of capsular melanocytic nevi compared with lymph nodes from patients with breast carcinoma.31 The higher percentage of patients whose lymph nodes were found to contain melanoma metastases by the PCR methods may be explained by the finding that those additional cases include a number of specimens with benign subcapsular melanocytic nevi. PCR techniques designed to detect tyrosinase may overdiagnose occult melanoma metastases, thereby leading to unnecessary therapy. In the Sunbelt Melanoma Trial currently in progress, one-fourth of the lymph node is taken for RT-PCR, while the other three-fourths is submitted for routine histology. If benign nevus cells are found histologically, the lymph node is declared negative, regardless of the RT-PCR results. This approach, while reducing the false-positive rate of the RT-PCR assays, does not address a potential problem in those cases in which RT-PCR is positive and neither nevus nor melanoma is identified histologically; nevus cells may have been present only in the one-fourth of the lymph node submitted for RT-PCR. On the other hand, PCR may have a significant role in diagnosing systemic metastases by testing patient's blood for the tyrosinase signal; benign nevus cells are not known to circulate in the blood stream. The effect of lymph node melanocytic nevi on serum tyrosinase levels remains unclear.
The use of antibodies to S-100 protein32 and NK1C333 has been shown to increase the rate of detection of occult melanoma metastases in regional lymph nodes. Antibodies to S-100 protein were found to be more sensitive than NK1C3, although S-100 antibodies also stained dendritic reticulum cells, capsular melanocytic nevus cells, sinus macrophages, and Schwann cells, which posed interpretative difficulties.32, 33 In addition, in a recent report of extensive sampling and staining of sentinel lymph nodes, Wen et al. did not report an increased detection rate over examination of a single hematoxylin and eosin–stained section using these techniques.34 On the other hand, some observers have reported an increased detection of microscopic metastases using immunohistochemical techniques.35
In our study, we found that combined step sectioning and immunohistochemistry were helpful in the identification of melanoma micrometastases in sentinel lymph nodes and that each of these two additional methods were equally important. Antibodies to MART-1 demonstrated intense staining of melanocytes without background staining. No cross-reaction with other cell types in the lymph nodes was noted in our study. Antibodies to S-100 protein were used mainly as a confirmatory stain. The use of both antibodies is recommended until more experience has been gained with the anti-MART-1 antibody. We found NK1C3 to be unhelpful due to its lack of specificity and high cost. Although HMB-45 may have a role in differentiating melanoma from benign melanocytic nevi in difficult cases, HMB-45 is not useful in the detection of melanocytic lesions in lymph nodes due to its low sensitivity. This finding is supported by recent results that HMB-45 positive melanoma cells are observed in 75% of lymph node metastases. Diffuse positive staining for HMB-45 is observed in only one-third of melanoma metastases.36 In summary, the value of anti-MART-1 antibody lies in its specificity; a positive signal serves to alert the pathologist to the presence of melanocytic cells in the lymph node, leading to more careful scrutiny of the adjacent hematoxylin and eosin–stained sections. If melanocytic atypia is detected, a diagnosis of microscopic melanoma metastasis can be made. The importance of identifying the tumor on an adjacent hematoxylin and eosin–stained section cannot be overemphasized, particularly in light of the results of our study and others reporting the frequency of lymph node nevi.37
In our study, we uncovered an additional 11.7% of sentinel lymph nodes harboring melanocytic micrometastases by performing 3 step sections through paraffin blocks covering, in total, a block thickness of approximately 0.25 mm. As the average tissue block is 2 mm thick, we may only have been “scratching the surface.” Wilkinson and Hause examined the probability of detecting micrometastasis of a certain size in a lymph node by examining one section from the central cut surfaces.38 They found that the probability of detecting a lesion of 0.5 mm (the average size of micrometastases in our study) in a lymph node of 10 mm in greatest dimension by 1 section through the center of the lymph node was 7.7%, assuming that the lesion could be located anywhere inside the lymph node with equal likelihood. Our study showed that micrometastases tended to be present mostly in the subcapsular sinuses (in 10 of 11 cases). It is possible that, were we to penetrate even more deeply into the tissue blocks, the number of positive cases may have increased, as supported by studies of the lymph nodes of patients with breast carcinoma.16, 17, 23
Notably, the detection rate in our study of 11.7% is entirely in keeping with the rates of 14% and 11% reported in previous studies of this type that employed slightly different methodology.14, 33 In the first of these studies, the question of whether more extensive sampling would yield an even higher frequency of lymph nodes containing occult tumor cells was addressed by examining an additional 9050 sections from 181 lymph nodes taken from 10 patients in whom the initial S-100 protein examination was negative with immunohistochemical staining of every 10th section. They did not find additional tumor cells in any of those additional sections. One explanation for this may be that metastatic melanoma cells preferentially deposit at the hilum of the lymph node, with the result that cross-sections through the center of the lymph node represent a good sample. As shown in Table 5, not all levels contained melanoma cells or nevus cells, indicating that deeper step sections do have a role in increasing the rate of detection of microscopic metastasis. The focal nature of the melanoma micrometastasis deposits suggest that frozen sections will reduce the sensitivity of subsequent analysis and should not be routinely performed on sentinel lymph node specimens. The trimming procedure at the time of frozen section wastes tissue and may cut through the only focus of microscopic metastases by the time the first section is taken. Instead, the lymph node should be properly fixed before careful cutting and sampling. The entire lymph node should be submitted for examination and cut into slices no thicker than 3 mm, then multiple deeper sections and immunohistochemical stains should be examined.
There were two cases that posed diagnostic difficulties even with the benefit of step sections and immunohistochemistry. In one case, two MART-1 positive cells were present within the lymph node capsule, but they could not be identified on the adjacent hematoxylin and eosin–stained sections. As a result, the morphology of these two cells could not be examined reliably, and the case was considered negative for metastases in the absence of other supporting features. The other case illustrated the difficulty of distinguishing between reactive sinus histiocytes and melanoma cells. In this case, some strongly NK1C3 positive cells were noted. The cells did not stain with antibodies to S-100, MART-1, or HMB-45. Examination of the adjacent hematoxylin and eosin–stained sections showed that the cells in question lacked nuclear pleomorphism and other features of malignancy, illustrating the value of examining adjacent hematoxylin and eosin–stained sections in difficult cases.
The prognostic significance of finding microscopic metastasis in a sentinel lymph node is still unclear. Whereas previous studies of breast carcinoma22, 24 showed that the presence of micrometastases did not imply a worse prognosis for the patient, data such as these are only beginning to emerge for melanoma. This uncertainty is compounded by the fact that most previous studies on melanoma metastatic to lymph nodes did not specify whether the melanoma deposits were macrometastases or micrometastases. More follow-up time is necessary to evaluate completely the utility of the sentinel lymph node biopsy procedure as a staging tool. The follow-up time in our study cohort was too short to allow us to correlate clinical outcome with pathologic analysis of the lymph nodes. Optimism regarding early results suggesting that adjuvant interferon α-2b improves the survival of patients with metastatic melanoma limited to regional lymph nodes (AJCC Stage III) supports the identification of these patients as early as possible in their disease course. However, recent trials of adjuvant interferon α-2b failed to show a significant difference in overall survival between patients receiving interferon treatment and controls.39 Because of the absence of any effective therapy for patients with widely metastatic melanoma, patients often will receive aggressive treatment at the first sign of metastasis. Sentinel lymph node analysis with microstaging allows for the earliest detection of metastatic melanoma.
This study demonstrates the utility of multiple sections and immunohistochemistry in the evaluation of sentinel lymph node specimens for metastatic melanoma. Previous studies that sought to determine whether the sentinel lymph node is a true reflection of the status of the rest of the lymph node basin were generally performed using the routine method of examining one or two hematoxylin and eosin–stained levels. Immunohistochemical stains for S-100 and HMB-45 were done, but only to confirm the identity of cytologically suspicious cells. As we have demonstrated, it is not always easy to determine the identity of suspicious cells solely with hematoxylin and eosin–stained sections. Indeed, in some cases positively staining cells were identified immunohistochemically, and with rereview of the hematoxylin and eosin–stained sections they were then recognized as abnormal in hindsight. We have now eagerly incorporated these techniques into the routine analysis of sentinel lymph node biopsies for melanoma. Specifically, we found that both the deeper levels and the immunohistochemical stains, in particular for S-100 and MART-1, were useful adjuncts to the routine methods of processing lymph nodes. Our current practice is to cut 3 sets of 3 serial sections at 80 μm intervals. The three slides in each set of serial sections are stained for S-100, hematoxylin and eosin, and MART-1, respectively.