1683

Sentinel Lymph Nodes in Malignant Melanoma Extended Histopathologic Evaluation Improves Diagnostic Precision

Helene Nortvig Abrahamsen, M.D.1 Stephen J. Hamilton-Dutoit, M.D.1 Jørn Larsen, M.D.2 Torben Steiniche, M.D.1 1

Institute of Pathology, Aarhus Kommunehospital, Aarhus University Hospital, Aarhus, Denmark.

2

Department of Plastic Surgery, Aarhus Kommunehospital, Aarhus University Hospital, Aarhus, Denmark.

BACKGROUND. The optimal technique for sentinel lymph node (SN) assessment in patients with melanoma is controversial. Molecular analysis (reverse transcriptase– polymerase chain reaction) detects significantly greater numbers of SNs with suspected micrometastases (up to 71%) than does routine histopathology (approximately 20%). The authors sought to identify possible reasons for this discrepancy and to determine whether using an extended histopathologic protocol could improve diagnostic precision. METHODS. Two hundred thirty-one SNs from 100 consecutive patients with cutaneous melanomas that measured 1– 4 mm in thickness were bisected, and half of the lymph node was examined according to an extensive histopathologic protocol involving serial sectioning and immunohistochemical analysis of 3 melanocyteassociated markers (S-100, HMB-45, and Melan-A). RESULTS. Lymph node melanocytic lesions were frequent, with micrometastases and benign nevus inclusions (BNI) found in SNs in 28% and 28% of patients, respectively (4 SNs contained both). Melan-A was the most sensitive immunohistochemical marker and was positive in all BNI-positive SNs and 97% of micrometastasis-positive SNs. Although HMB-45 showed differential labeling in micrometastases compared with BNI (82% vs. 16%), immunohistochemistry could not distinguish between those lesions. Micrometastases were already identified on the first central level in 49% of positive SNs, whereas only 23% of SNs with BNI were diagnosed on the first level. CONCLUSIONS. Extensive serial sectioning with immunohistochemical analysis substantially increased the histopathologic detection of micrometastases and BNI in melanoma SNs to a level approaching the level reported for molecular techniques. The large number of BNIs represents an important potential source of imprecision (false positivity) in SN assays based on nonmorphologic methods. Cancer 2004;100:1683–91. © 2004 American Cancer Society. KEYWORDS: melanoma, sentinel lymph nodes, pathology, immunohistochemistry.

Supported by the Institute for Experimental Clinical Research at Aarhus University; the Clinical Research Unit of the Danish Cancer Society; the Forskningsinitiativet of the County of Aarhus; Aage Bang’s Foundation; the Danish Cancer Research Foundation; Erland Richard Frederiksen’s Legate; Fritz, Georg, and Marie Cecilie Glud’s Foundation; and Meta and Haakon Bagger’s Foundation. Address for reprints: Helene Nortvig Abrahamsen, M.D., Institute of Pathology, Aarhus Kommunehospital, Noerrebrogade 44, DK-8000 Aarhus C, Denmark; Fax: (011) 45 89493690; E-mail: [email protected] Received August 15, 2003; revision received February 3, 2004; accepted February 3, 2004.

T

he most important prognostic factor for survival in patients with cutaneous malignant melanoma is the Breslow tumor thickness.1 However, once the patient develops lymph node metastases, the histopathologic features of the primary tumor are no longer useful for predicting survival.2 This makes early and precise assessment of eventual regional lymph node involvement important. The sentinel lymph node (SN) strategy is a promising tool for this purpose.3 The procedure involves excision and pathologic examination of the first tumor draining lymph node, the site at which initial metastasis is most likely to occur. Conceptually, a negative SN predicts with high accuracy the absence of metastatic tumor in the remaining lymph node basin. SN status has an important impact on prognosis for patients with melanoma and is now included in the revised staging system for cutaneous

© 2004 American Cancer Society DOI 10.1002/cncr.20179 Published online 18 March 2004 in Wiley InterScience (www.interscience.wiley.com).

1684

CANCER April 15, 2004 / Volume 100 / Number 8

melanoma proposed by the American Joint Committee of Cancer.2 Whether using this technique also will lead to better patient survival is being studied currently (in the Sunbelt Melanoma Trial).4 The SN procedure is suited well to improve detection of subclinical tumors. Because only limited numbers of lymph nodes are removed, it is possible to perform a more extensive histopathologic search for small metastatic deposits using both serial sectioning and immunohistochemical staining. However, although it is apparent that the ability to detect lymph node micrometastases is dependent on the histopathologic protocol used, the best way to examine the SN is controversial. Thus, the number of levels and sections from the SN to be examined and the need for and choice of immunohistochemical staining remain unclear.5–7 Recently, there has been considerable interest in the use of molecular biologic assays as alternative methods for assessing SN status. For example, the reverse transcriptase–polymerase chain reaction (RTPCR) can be used to analyze SN tissue extracts for mRNA from melanocyte-associated genes, such as tyrosinase and MART1.8 –12 In general, these assays appear to be capable of detecting melanocytic cells in SNs with a significantly greater sensitivity (up to 71%)13 compared with routine histopathology (approximately 20%). The discrepancy between histopathology and RT-PCR suggests that the former may underestimate the true incidence of micrometastases, presumably because, for practical reasons, histology actually examines ⬍ 1% of the submitted material. However, most molecular studies of SN status are based on detection of mRNA sequences expressed by both benign and malignant melanocytes. Because SNs may contain benign nevus cell inclusions, molecular assays of whole lymph node extracts may overestimate the true incidence of melanoma metastases. The objective of the current study was to investigate possible reasons for the discrepancy in incidence of melanoma metastases in SNs detected by histologic and molecular methods. In particular, we wanted to know whether using an extensive histopathologic protocol, consisting of serial sectioning with immunohistochemical staining for three melanocyte-associated markers (S-100, HMB-45, and Melan-A), would increase the number of melanoma metastases and benign nevus cell inclusions detected.

MATERIALS AND METHODS Patients From January 2001 to December 2002, 100 patients age ⬍ 71 years with intermediate-thickness (1– 4 mm) primary cutaneous melanomas underwent staging of

the regional lymph node basin by SN dissection at our hospital. None of the patients showed clinical evidence of metastatic disease in regional lymph nodes or at distant sites. Written informed consent based on a protocol approved by the local ethical authorities was obtained from all patients.

SN Biopsy Procedure Lymphatic drainage patterns were established using previously published methods14,15 with minor modifications. In brief, approximately 24 hours before surgery, a total dose of 80 megabecquerels of 99mtechnetium-labeled nanocolloid (Nycomed Amersham Sorin, Saluggia, Italy) was injected at 4 sites in the dermis surrounding the scar of the previously excised melanoma. Planar gamma images were taken in all patients within 2 hours of injection and were used to identify the draining lymphatic regions. A hand-held gamma detector (C-Trak Navigator; Morgan Hill, CA) was used intraoperatively to identify radioactive lymph nodes. In the operating theater, the excised lymph nodes were trimmed of fat and bisected along their long axes, with the plane of sectioning passing as close to the hilum as possible (based on macroscopic assessment). In many smaller lymph nodes, the hilum was difficult to identify precisely. The gamma probe was used to confirm that there was radioactivity in both halves. Half of the lymph node was fixed in neutral buffered formaldehyde, and the other half was snap-frozen in liquid nitrogen for subsequent molecular analysis.

Histologic and Immunohistochemical Processing The fixed lymph node tissue for histopathology was embedded in paraffin with the (central) cut surface oriented for sectioning. Lymph nodes that measured ⬎ 4 mm were divided parallel to their long axis into 2–3 mm slices, each of which was embedded separately. Each paraffin block was sectioned entirely according to an extensive histopathologic protocol. Levels were cut at 250 ␮m intervals; sections were 2 ␮m thick. One hematoxylin and eosin (H & E) section was stained at each level (i.e. Levels 1, 2, 3, 4, etc.). At alternate levels, starting with the first (i.e. Levels 1, 3, 5, 7, etc.), an additional 3 sections were cut for immunohistochemical staining with primary antibodies against S-100 protein (MU058-UC; dilution, 1:10,000; BioGenex, Sakura Prohosp, Vaerlose, Denmark), gp100 (HMB-45; dilution, 1:50; M0634; DakoCytomation, Glostrup, Denmark), and Melan-A (M7196; dilution, 1:50; DakoCytomation). To maximize consistency, immunohistochemistry was performed using an automated immunostainer (Techmate 500; DakoCytomation) according to the following protocol: 2 ␮m paraffin sections were mounted on ChemMate slides

Pathology in Melanoma Sentinel Lymph Nodes/Abrahamsen et al. TABLE 1 Pathologic Criteria for Melanoma Micrometastases in Sentinel Lymph Nodes Micrometastases 1. Clinically nonpalpable 2. Located in subcapsular sinus and/or lymph node parenchyma 3. Exhibit cellular atypia: cell enlargement, nuclear pleomorphism, prominent nucleoli; atypical cells identified in hematoxylin and eosin–stained section 4. Exhibit positive immunohistochemical staining for at least one melanocytic marker (S-100, HMB-45, Melan-A)

(DakoCytomation) and dried for 1 hour at 60 °C. Sections were deparaffinized, rehydrated through graded ethanol, and washed in tap water. Antigen retrieval was performed by microwave heating in a TEG buffer (Tris-ethylene glycol tetraacetic acid, pH 9.0) for 25 minutes. Sections were cooled to room temperature, placed in Tris-buffered saline (TBS) (50 mM Tris/150 mM NaCl, pH 7.6), and then incubated with the primary antibody diluted in ChemMate antibody diluent (S2022; DakoCytomation) for 60 minutes before rinsing 3 times in TBS for 3 minutes each. Endogenous peroxidase was blocked by incubation in methanol and H2O2 for 10 minutes followed by washing in tap water for 2 minutes. Bound primary antibody was detected using horseradish peroxidase Envision polymer (K 4001; DakoCytomation) for 30 minutes followed by rinsing 3 times in TBS for 3 minutes each. Bound second-layer polymers were visualized by incubation with the chromogen diaminobenzidine tetrahydrochloride for 10 minutes followed by rinsing in TBS for 2 minutes. The reaction product was enhanced by incubation with a CuSO4 (C-7631; Sigma, St. Louis, MO) solution for 5 minutes followed by a 2-minute rinse in tap water. Finally, sections were counterstained with hematoxylin and mounted in Aquatex (64912-50; Kebo-Lab, Albertslund, Denmark). Appropriate positive controls were included for each stain. All slides were examined by an experienced pathologist with an interest in melanoma. Unusual findings were examined by at least two pathologists, and a consensus diagnosis was reached. The patient was considered to have a positive SN when all the criteria listed in Table 1 were fulfilled. To minimize the possibility of misinterpreting benign nevus cells as melanoma, we deliberately chose rigorous criteria for defining metastasis. These included a requirement that atypical cells should be identified microscopically in an adjacent H & E section in all lymph nodes in which suspect cells were found on immunostaining. Although this policy may have led to a slight underestimation of the number of metastases, we believe that this cautious approach was justified for reducing the

1685

number of false-positive diagnoses of micrometastases. In the current study, micrometastasis was defined as a clinically nonpalpable lymph node containing melanoma, as defined in Table 1, irrespective of the size of the metastasis.

Statistical Analysis Comparisons of mean primary tumor thickness in patients with metastasis-positive SNs, benign nevus inclusion–positive SNs, and melanocyte-free SNs were performed using Kruskal–Wallis and Mann–Whitney tests. Chi-square tests were used to compared the incidence of micrometastases.

RESULTS Patient and Primary Tumor Characteristics One hundred patients were included in the study. Five patients did not participate in the molecular part of the project. Thus, their lymph nodes were bisected and examined in toto according to the histopathologic procedure described above. The clinical details of the patients and the pathologic features of their tumors are shown in Table 2. Tumor thickness could not be determined in 14 patients, either because the primary tumor had been cut through or because of substantial loss of the primary tumor due to ulceration or regression.

SNs In total, 231 SNs were excised and studied; at least 1 SN was harvested successfully in each patient. The mean and median SN numbers per patient were 2.3 and 2.0, respectively (range, 1– 6 SNs). One SN was harvested in 26 patients, 2 SNs were harvested in 39 patients, 3 SNs were harvested in 21 patients, and ⬎ 3 SNs were harvested in 14 patients.

Histopathology Micrometastases SN micrometastases were identified in 28 patients (28%). Four of those patients also had SNs with benign nevus inclusions (Fig. 1), 20 had 1 positive SN, 5 had 2 positive SNs, and 3 had 3 positive SNs (median, 1.0 positive SNs). Micrometastases were found either as small clusters or as larger, solid areas of melanoma cells, or they were dispersed in the lymph node as single cells. In 39 tumor-positive SNs, metastases were located in the subcapsular sinus (18 of 39 SNs; 46%), in the lymph node parenchyma (7 of 39 SNs; 18%), or at both sites (13 of 39 SNs; 33%). In 1 SN (3%), there was a single, small, intravascular focus of melanoma related to the fibrous trabeculae. Primary melanomas in patients who had metastasis-positive SNs were significantly thicker compared with primary melanomas in

1686

CANCER April 15, 2004 / Volume 100 / Number 8

TABLE 2 Tumor Pathology and Clinical Details in 100 Patients who Underwent Sentinel Lymph Node Analysis Feature Gender Female Male Age (yrs) Mean Range Histologic type SSM NM ALM DMM Unclassified Breslow thickness 1.0–2.0 mm 2.1–3.0 mm 3.1–4.0 mm Unclassified Mean (mm) Median (mm) Clark level III IV V Unclassified Ulceration Absent Present Unclassified Primary site Extremities Trunk Head and neck

No. of patients

53 47 50.5 26–70 61 28 1 1 9 60 17 9 14 1.84 1.56

FIGURE 1. Results of histopathologic examination of sentinel lymph nodes from 100 patients with melanoma. Percentages refer to patient incidence rates. BNI: benign nevus inclusions; Free: melanocyte-free; Both: BNI ⫹ metastases.

38 51 1 10 75 23 2 40 55 5

SSM: superficial spreading melanoma; NM: nodular melanoma; ALM: acrolentiginous melanoma; DMM: desmoplastic malignant melanoma.

FIGURE 2. Sentinel lymph node (SN) status versus tumor Breslow thickness patients who had SNs that contained no melanocytes (mean thickness, 2.10 mm vs. 1.64 mm) (Fig. 2). The mean tumor thickness in patients who had SNs that contained benign nevus inclusions was intermediate (1.94 mm) and did not differ significantly from the tumor thickness in patients who had metastasis-positive SNs or from the tumor thickness in patients who had SNs without melanocytes.

Benign nevus inclusions Benign nevus inclusions were identified in 28 patients (28%), including 4 who also had micrometastases (Fig. 1); thus, they were found in 14% of patients with micrometastases and in 33% of patients without micrometastases. Benign nevus inclusions were found in 31 of 231 SNs (13.4%), and the majority of patients only had inclusions in a single SN. Nevus inclusions

of the corresponding primary melanoma. Mean tumor thicknesses are indicated by horizontal lines. The mean tumor thickness was significantly greater in metastasis-positive SNs (2.10 mm) compared with melanocyte-free SNs (1.64 mm; P ⫽ 0.0002). Mean tumor thickness in the benign nevus inclusion (BNI)-positive SNs was intermediate (1.94 mm) and did not differ significantly from the corresponding figure in metastatic SNs (P ⫽ 0.22) or melanocyte-free SNs (P ⫽ 0.26). Free: melanocyte-free.

appeared as small nests of epithelioid melanocytes located exclusively in the lymph node capsule (71%), in the fibrous trabeculae entering the lymph node (19%), or in the lymph node parenchyma (3%; 1 patient). The location of this single parenchymal nevus was confirmed on staining a parallel section with Masson trichrome. Two patients (7%) had benign nevus inclusions in both the lymph node capsule and the fibrous trabeculae. We found no evidence of blue nevus cells in the lymph nodes in this series.

Pathology in Melanoma Sentinel Lymph Nodes/Abrahamsen et al. TABLE 3 Immunohistochemical Profile of Micrometastases and Benign Nevus Inclusions in Sentinel Lymph Nodes No. of sentinel lymph nodes (%) Diagnosis

H&E

S-100

HMB-45

Melan-A

Micrometastases (n ⫽ 39)a BNI (n ⫽ 31)

39 (100) 15 (48)

35 (90) 22 (71)

32 (82) 5 (16)

38 (97) 31 (100)

BNI: benign nevus inclusion; H & E: hematoxylin and eosin. a By definition, all metastases are identified on hematoxylin and eosin–stained slides.

1687

levels distributed throughout the entire lymph node. In metastasis-positive SNs, melanoma cells could be identified by immunohistochemical markers in an average of 72% of the levels. Benign nevus inclusions did not appear to be overrepresented in the central part of the SNs. Thus, inclusions in positive SNs were identified after examination of the first central level in only 7 of 31 SNs (23%) (Fig. 4). Nevus inclusions frequently were located at a single level, often as a solitary group of a few cells. In inclusion-positive SNs, nevus cells were present in an average of 37% of the levels.

DISCUSSION Immunohistochemical profile Micrometastases and benign nevus inclusions often were identified first in the immunohistochemical sections, typically with the Melan-A stain. By definition, metastases always were identified subsequently on the H & E sections. In contrast, benign nevus inclusions could be found on routinely stained sections in only approximately one-half of the sections examined. Positivity for Melan-A was found in 69 of 70 melanocytic lesions (97%) (Table 3). In contrast, the melanocyteassociated marker S-100 was identified in association with 35 of 39 micrometastases (90%) and 22 of 31 benign nevus inclusions (71%). This apparent discrepancy was attributed to difficulties in distinguishing small groups of melanocytic cells from the many other S-100-positive cells that normally are present in and around reactive lymph nodes. Staining with HMB-45 differed significantly when comparing metastases with benign nevus inclusions: 32 of 39 metastases were positive (82%), whereas only 5 of 31 benign nevus inclusions were positive (16%). We found no difference in staining intensity comparing HMB-45-positive nevus cells with melanoma cells (Fig. 3).

Multilevel sectioning Our protocol included extensive serial sectioning (thickness, 250 ␮m) of the half of the SN that was used for histopathology. The median number of levels cut per SN was 8 (range, 4 –23 levels), and the median number of H & E and immunohistochemical sections examined per SN was 20 (range, 10 –59 sections). We defined the first four sections (one stained with H & E and three immunostained) cut from the bisected face of the lymph node as the first central level. A diagnosis of micrometastasis was made correctly after microscopy of this first central level in 19 of 39 positive SNs (49%) and after microscopy of the first 3 levels (corresponding to a centrally located region 500 ␮m in thickness) in 31 of 39 positive SNs (79%) (Fig. 4). However, micrometastases were found not only in sections from the central portion of the SNs but also often at several

We used an extended histopathologic protocol for assessing SNs that combined extensive serial sectioning with multiple immunohistochemical stains. This protocol greatly increased the detection of both benign nevus inclusions and melanoma metastases compared with previously published studies. Overall, melanocytic lesions were identified in SNs from 52 of 100 patients (52%). Based on strict diagnostic criteria, these findings were classified as metastases in 24 patients (24%), benign nevus inclusions in 24 patients (24%), and both metastases and nevus inclusions in 4 patients (4%). Thus, the overall incidence of SN metastases of 28%, was significantly greater compared with what was found in previous histopathologic studies, which reported rates of 14 –22%,3,8,16 –20 despite the finding in the current study that the mean tumor thickness (1.8 mm) was comparable to or even less than the mean tumor thickness values reported in earlier studies.13,17,20 Similarly, benign lymph node inclusions were more common in the current study than previously assumed, with a patient incidence of 28% and a lymph node (SN) incidence as high as 13.4%. Benign nevocytes have been reported in lymph nodes from melanoma patients in a number of studies, with a patient incidence of some 3– 6% and a lymph node incidence of 0.12– 4.0%.21–23 Carson et al.24 examined lymph node dissection specimens following SN analysis in 208 patients with melanoma. Analyzing a total of 4821 lymph nodes, those authors found a 1.2% lymph node incidence of benign nevus inclusions, corresponding to an SN incidence of 3.9%, and an overall patient incidence of 23.6%. In contrast, in control material from over 1500 lymph nodes that were removed from patients with breast and pelvic carcinomas, those authors detected benign inclusions in only 1 lymph node (0.06%), indicating that lymph node nevocytes are selectively present in patients with melanoma. We believe that the higher incidence of benign and malignant melanocytes found in the current study can be attributed primarily to two factors: 1) the use of

1688

CANCER April 15, 2004 / Volume 100 / Number 8

FIGURE 3. Sentinel lymph nodes from patients with malignant melanoma. (A) A benign nevus inclusion is seen in the capsule of the lymph node. The inset shows a higher-power view of the nevus cells. (B) These cells exhibit positive staining for the melanocytic marker HMB-45. (C) A lymph node sample from another patient shows micrometastatic melanoma in the subcapsular sinus, which is seen at higher power in the inset. (D) These melanoma cells exhibit positive staining for HMB-45 (D).

FIGURE 4. Data on correct sentinel lymph node assessment according to the number of levels examined. The number of examined levels that were required to identify a given percentage of melanoma metastases or benign nevus inclusions (BNI) in sentinel lymph nodes (SN) is shown. Levels were cut at 250 ␮m intervals. Level 1 is the central level, and Level 11 the peripheral level. Staining with hematoxylin and eosin (H & E), S-100, HMB-45, and Melan-A was performed on Levels 1, 3, 5, 7, 9, and 11. H & E staining only was performed on Levels 2, 4, 6, 8, and 10. Calculations were based on the number of SNs with micrometastases, BNI, or both. Thirty-nine SNs with micrometastases and 31 SNs with BNI were investigated.

extensive serial sectioning; and 2) the use of multiple immunohistochemical stains. In general, the histopathologic protocols used in previous SN studies have not been described well. In several studies, the pathologic examination appeared to consist of ‘bivalving’ the SN and staining a single central section with H & E, sometimes supplemented by one or more immu-

nostains. Surprisingly, specific details of the type and number of sections are not even included in several reports.12,17,25 In studies in which details of the histologic analysis have been described, protocols vary but usually involve examination of a single H & E section and one or two immunohistochemical stains (often with S-100 and/or HMB-45) restricted to the central first to third levels, resulting in an average of nine sections examined.8,9,13,18 –20,26 –28 Although we used only half of each SN for histologic study, the tissue was sectioned completely at 250 ␮m intervals. Each level was stained with H & E;, and at alternating levels, we performed immunohistochemical staining with S-100, HMB-45, and Melan-A, resulting in an average of 20 sections from 8 different levels in each SN. Thus, a much larger proportion of the SN tissue was studied microscopically using the current protocol compared with most previous studies. Although it is possible that we would have found even more melanocyte-positive lymph nodes if the other half of the specimen had been subjected to the same rigorous pathologic study, we believe that our incidence figures for metastases and lymph node nevi are more reliable compared with the results from most previous studies, which have been based on a limited sampling of SNs. We found that the majority of micrometastases were located in the subcapsular/intermediate sinuses,

Pathology in Melanoma Sentinel Lymph Nodes/Abrahamsen et al.

and benign nevus inclusions were located in the fibrous lymph node capsule, as reported previously.24 It is noteworthy that by bisecting the SN, we were able to diagnose 31 of 39 tumor-positive SNs (79%) correctly after examination of the first 3 levels (Fig. 4), corresponding to a centrally located region 500 ␮m in thickness. Cochran et al.29 have argued that melanomas typically metastasize first to the subcapsular space corresponding to a central plane through the hilum and the longest dimension of the lymph node. This is consistent with our finding that metastases appeared to be located selectively in the central part of the lymph node, and it emphasizes the importance of accurate bisection of the SN to ensure efficient histopathologic examination. Benign nevus inclusions did not show the same relation to the midline of the lymph nodes (Fig. 4). Only 14 of 31 inclusion-positive SNs (45%) were diagnosed correctly after examination of the first 3 levels. The sporadic distribution of nevus inclusions, as well as the fact that they often appeared as a single focus or just as a cluster of few cells in the periphery of the SN, also helps explain why their incidence has been underestimated in previous studies. The current protocol also differed from protocols used in previous studies in its use of several immunohistochemical stains on alternate levels after serial sectioning. We chose the panel of immunohistochemical markers based on their sensitivity and specificity for melanocytic cells. In addition, we were interested in comparing the use of Melan-A with the more traditional melanoma markers S-100 and HMB-45. S-100 is a sensitive but not very specific marker of melanocytes.19,24,30 In lymph nodes, S-100 immunoreactivity will be seen not only in melanocytes but also in activated macrophages, small nerve twings, adipocytes, and dendritic cells.31 It can be very difficult to identify small foci of melanocytic cells against this distracting background, as demonstrated in the current study, in which only 71% of lymph nodes that contained benign nevus inclusions and 90% of lymph nodes with metastases were identified using S-100-immunostained sections (Table 3). Compared with S-100, HMB-45 is a relatively specific marker for melanocytes that lacks sensitivity.19 We found that 18% of lymph nodes with micrometastases were HMB-45-negative. Other studies have found that up to 30 –35% of metastases were negative for HMB-45.31,32 HMB-45 preferentially stains activated or immature melanocytes.33 Thus, it is not surprising that several studies have reported that benign nevus inclusions (which can be considered nonactivated melanocytes) were negative for this marker.5,31,32 Indeed, HMB-45 has been used to distinguish such inclusions from micrometastases. However, we found

1689

HMB-45 positivity in 16% of the nevus-positive lymph nodes in this study. None of those HMB-45-positive inclusions showed cellular atypia, and all but one were found in the fibrous capsule. Thus, our data indicate that HMB-45 alone cannot be used to distinguish between micrometastases and benign nevus inclusions. This observation is supported by Lohman et al.,34 who found focal, positive HMB-45 staining in 2 of 15 lymph node benign nevus inclusions (13%). Melan-A is a recently introduced immunohistochemical marker that reportedly is expressed in most primary and metastatic malignant melanomas.35 Positive reactions also are seen in several nonmelanocytic tumors (e.g., angiomyolipomas, adrenal cortical tumors, and sex-cord stromal tumors), but these are not relevant in the setting of SN analysis.36,37 Melan-A has sensitivity for melanocytic cells similar to that of S-100, although Melan-A staining in desmoplastic melanomas is variable, and the correct diagnosis of this lesion (and its metastases) requires detailed histologic examination and the use of S-100 immunostaining.35 In agreement with Shidham et al.,31 we found no background or nonmelanocyte staining using the Melan-A antibody in SN sections. This combination of high sensitivity and specificity makes MelanA an ideal screening tool that helps identify even single melanocytes within lymph nodes. We used Melan-A staining (in combination with S-100 staining) to alert the pathologist to the presence of melanocytes in the SN, which could then be categorized further as micrometastases or benign nevus inclusions based on their morphology on H & E–stained sections. Despite using our strict pathologic guidelines (Table 1), such differentiation was not always straightforward. Previous large SN studies (n ⬎ 60) have used the traditional immunohistochemical markers S-100 and HMB-45 rather than Melan-A. We found that Melan-A was a very helpful marker in SN analysis, and we believe that its use in our study, in combination with S-100 and HMB-45, provides another explanation for the significantly greater incidence of micrometastases found compared with earlier reports. Similarly, the relatively high number of lymph node benign nevus inclusions found in the current study compared with SN studies in which Melan-A staining has not been performed21,23,24 is not surprising. These cell clusters, which frequently are quite small, often are HMB-45negative and are difficult to identify using S-100 staining against a background of numerous S-100-positive nonmelanocytic cells. Does the high frequency of benign nevus inclusions found in the current study have any clinical implications? It appears that patients with melanoma have a higher incidence of benign nevus inclusions in

1690

CANCER April 15, 2004 / Volume 100 / Number 8

lymph nodes compared with patients who have breast and pelvic carcinomas.23,24 This may be explained in part by the fact that SNs from patients with melanoma (but not SNs from patients with breast carcinoma) are skin-draining lymph nodes: The hypothesis is that lymph node nevocytes are derived from cutaneous nevi. However, comparing lymph node dissection specimens that include both skin-draining and nonskin-draining lymph nodes, patients with melanoma still appear to have a higher incidence of lymph node nevus inclusions compared with patients who have other types of malignant disease.24 Could the presence of benign nevus inclusions be a negative prognostic indicator? If so, then the expectation would be to find benign nevus inclusions with greater frequency in patients with metastases than in patients without them. This was not the case. In fact, only a minority of patients (14%) with metastatic SN also had benign nevus inclusions. One aspect of SN evaluation for which the relatively high frequency of benign nevus inclusions already is important is in the use of molecular tools to detect metastases. Molecular analysis is based on the PCR-based detection of mRNA derived from melanocyte-associated genes, such as tyrosinase and MART1, in tissue extracts from SNs. Using molecular analysis, melanocyte-associated mRNA can be detected in 44 – 71% (mean, 56%) of SNs.8,9,11–13 It is noteworthy that this is not very different from the frequency of positive SNs found in the current study (52%). Naturally, our results cannot be related directly to the molecular rates for positive SNs, because they are not derived from the same material. However, they do suggest that extensive histopathologic analysis may increase the sensitivity of detection of melanocytes in SNs to a level similar to that possible with molecular analysis. However, an important difference is that in the current morphologic analysis, we were able to identify the melanocytes found in nearly one-half of positive SNs as benign nevus inclusions (24%). Because tyrosinase and MART1 mRNA are expressed in both benign and malignant melanocytes,38 a high frequency of nevus inclusions may lead to a high false-positive rate with consequent risk of overtreatment if molecular analysis alone is used for SN assessment. Does the high frequency of metastases found in our study have any clinical implications? We found approximately 40% more tumor-positive SNs than were found in previous reports. In particular, the use of Melan-A immunohistochemistry meant we could detect micrometastases that consisted of only a few cells. It is possible that such small metastases may be unlikely to progress and thus may be of no clinical relevance; this remains to be investigated. The pres-

ence or absence of micrometastases appears to be a key prognostic marker.2 However, there also is evidence that the amount of metastatic melanoma in a positive SN may be an independent prognostic factor.28,39 Although there is no direct evidence that very small micrometastases are of clinical importance, it is interesting to note that our incidence rate for tumorpositive SNs (28%) is correlated quite well with clinical observations, which show that almost one-third of patients with melanoma treated by excision alone will go on to develop regional lymph node metastases on clinical follow-up.40,41 In conclusion, the current study shows that the detection of benign and malignant melanocytes in SNs is related to the extent of the histopathologic and immunohistochemical examination. Our results— especially the finding of a high frequency of benign nevus inclusions in SNs—provide an explanation for the apparent discrepancies in the results from previous studies using histologic methods compared with molecular methods of SN analysis. Our findings have important implications for the planning of future protocols for pathologic analysis of SNs. Although SN analysis will always involve a compromise between optimal sampling and examination of the entire lymph node and the practical and economic limitations of routine practice, the current study shows that intensive histopathologic examination will improve the precision of SN analysis.

REFERENCES 1.

2.

3.

4.

5.

6.

7.

8.

Breslow A. Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma. Ann Surg. 1970;172:902–908. Balch CM, Buzaid AC, Soong SJ, et al. Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol. 2001;19:3635–3648. Morton DL, Wen DR, Wong JH, et al. Technical details of intraoperative lymphatic mapping for early stage melanoma. Arch Surg. 1992;127:392–399. McMasters KM, Sondak VK, Lotze MT, Ross MI. Recent advances in melanoma staging and therapy. Ann Surg Oncol. 1999;6:467– 475. Lawrence WD. ADASP recommendations for processing and reporting of lymph node specimens submitted for evaluation of metastatic disease. Virchows Arch. 2001;439:601– 603. Prieto VG, Clark SH. Processing of sentinel lymph nodes for detection of metastatic melanoma. Ann Diagn Pathol. 2002; 6:257–264. van Diest PJ, Peterse HL, Borgstein PJ, Hoekstra O, Meijer CJ. Pathological investigation of sentinel lymph nodes. Eur J Nucl Med. 1999;26:S43–S49. Blaheta HJ, Ellwanger U, Schittek B, et al. Examination of regional lymph nodes by sentinel node biopsy and molecular analysis provides new staging facilities in primary cutaneous melanoma. J Invest Dermatol. 2000;114:637– 642.

Pathology in Melanoma Sentinel Lymph Nodes/Abrahamsen et al. 9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

Bostick PJ, Morton DL, Turner RR, et al. Prognostic significance of occult metastases detected by sentinel lymphadenectomy and reverse transcriptase-polymerase chain reaction in early-stage melanoma patients. J Clin Oncol. 1999; 17:3238 –3244. Wang X, Heller R, VanVoorhis N, et al. Detection of submicroscopic lymph node metastases with polymerase chain reaction in patients with malignant melanoma. Ann Surg. 1994;220:768 –774. Gutzmer R, Kaspari M, Brodersen JP, et al. Specificity of tyrosinase and HMB45 PCR in the detection of melanoma metastases in sentinel lymph node biopsies. Histopathology. 2002;41:510 –518. Shivers SC, Wang X, Li W, et al. Molecular staging of malignant melanoma: correlation with clinical outcome. JAMA. 1998;280:1410 –1415. Li W, Stall A, Shivers SC, et al. Clinical relevance of molecular staging for melanoma: comparison of RT-PCR and immunohistochemistry staining in sentinel lymph nodes of patients with melanoma. Ann Surg. 2000;231:795– 803. Bongers V, Borel RI, Barneveld PC, Canninga-van Dijk MR, van Rijk PP, van Vloten WA. Towards quality assurance of the sentinel node procedure in malignant melanoma patients: a single institution evaluation and a European survey. Eur J Nucl Med. 1999;26:84 –90. Valdes Olmos RA, Hoefnagel CA, Nieweg OE, et al. Lymphoscintigraphy in oncology: a rediscovered challenge. Eur J Nucl Med. 1999;26:S2–S10. Cascinelli N, Belli F, Santinami M, et al. Sentinel lymph node biopsy in cutaneous melanoma: the WHO Melanoma Program experience. Ann Surg Oncol. 2000;7:469 – 474. Clary BM, Brady MS, Lewis JJ, Coit DG. Sentinel lymph node biopsy in the management of patients with primary cutaneous melanoma: review of a large single-institutional experience with an emphasis on recurrence. Ann Surg. 2001;233: 250 –258. Gershenwald JE, Thompson W, Mansfield PF, et al. Multiinstitutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 Stage I or II melanoma patients. J Clin Oncol. 1999;17:976 –983. Messina JL, Glass LF, Cruse CW, Berman C, Ku NK, Reintgen DS. Pathologic examination of the sentinel lymph node in malignant melanoma. Am J Surg Pathol. 1999;23:686 – 690. Morton DL, Thompson JF, Essner R, et al. Validation of the accuracy of intraoperative lymphatic mapping and sentinel lymphadenectomy for early-stage melanoma: a multicenter trial. Multicenter Selective Lymphadenectomy Trial Group. Ann Surg. 1999;230:453– 463. Baisden BL, Askin FB, Lange JR, Westra WH. HMB-45 immunohistochemical staining of sentinel lymph nodes: a specific method for enhancing detection of micrometastases in patients with melanoma. Am J Surg Pathol. 2000;24:1140 – 1146. McCarthy SW, Palmer AA, Bale PM, Hirst E. Naevus cells in lymph nodes. Pathology. 1974;6:351–358. Ridolfi RL, Rosen PP, Thaler H. Nevus cell aggregates associated with lymph nodes: estimated frequency and clinical significance. Cancer. 1977;39:164 –171. Carson KF, Wen DR, Li PX, et al. Nodal nevi and cutaneous melanomas. Am J Surg Pathol. 1996;20:834 – 840.

1691

25. Boi S, Cristofolini P, Togni R, et al. Detection of nodal micrometastases using immunohistochemistry and PCR in melanoma of the arm and trunk. Melanoma Res. 2002;12: 147–153. 26. Essner R, Conforti A, Kelley MC, et al. Efficacy of lymphatic mapping, sentinel lymphadenectomy, and selective complete lymph node dissection as a therapeutic procedure for early-stage melanoma. Ann Surg Oncol. 1999;6:442– 449. 27. Starz H, De Donno A, Balda BR. The Augsburg experience: histological aspects and patient outcomes. Ann Surg Oncol. 2001;8:48S–51S. 28. Starz H, Balda BR, Kramer KU, Buchels H, Wang H. A micromorphometry-based concept for routine classification of sentinel lymph node metastases and its clinical relevance for patients with melanoma. Cancer. 2001;91:2110 –2121. 29. Cochran AJ, Roberts AA, Saida T. The place of lymphatic mapping and sentinel node biopsy in oncology. Int J Clin Oncol. 2003;8:139 –150. 30. Cochran AJ, Lu HF, Li PX, Saxton R, Wen DR. S-100 protein remains a practical marker for melanocytic and other tumours. Melanoma Res. 1993;3:325–330. 31. Shidham VB, Qi DY, Acker S, et al. Evaluation of micrometastases in sentinel lymph nodes of cutaneous melanoma: higher diagnostic accuracy with Melan-A and MART-1 compared with S-100 protein and HMB-45. Am J Surg Pathol. 2001;25:1039 –1046. 32. Yu LL, Flotte TJ, Tanabe KK, et al. Detection of microscopic melanoma metastases in sentinel lymph nodes. Cancer. 1999;86:617– 627. 33. Skelton HG III, Smith KJ, Barrett TL, Lupton GP, Graham JH. HMB-45 staining in benign and malignant melanocytic lesions. A reflection of cellular activation. Am J Dermatopathol. 1991;13:543–550. 34. Lohmann CM, Iversen K, Jungbluth AA, Berwick M, Busam KJ. Expression of melanocyte differentiation antigens and Ki-67 in nodal nevi and comparison of Ki-67 expression with metastatic melanoma. Am J Surg Pathol. 2002;26:1351–1357. 35. Orosz Z. Melan-A/Mart-1 expression in various melanocytic lesions and in non-melanocytic soft tissue tumours. Histopathology. 1999;34:517–525. 36. Busam KJ, Jungbluth AA. Melan-A, a new melanocytic differentiation marker. Adv Anat Pathol. 1999;6:12–18. 37. Stewart CJ, Nandini CL, Richmond JA. Value of A103 (Melan-A) immunostaining in the differential diagnosis of ovarian sex cord stromal tumours. J Clin Pathol. 2000;53: 206 –211. 38. Messina JL, Glass LF. Pathologic examination of the sentinel lymph node. J Fla Med Assoc. 1997;84:153–156. 39. Carlson GW, Murray DR, Lyles RH, Staley CA, Hestley A, Cohen C. The amount of metastatic melanoma in a sentinel lymph node: does it have prognostic significance? Ann Surg Oncol. 2003;10:575–581. 40. Meier F, Will S, Ellwanger U, et al. Metastatic pathways and time courses in the orderly progression of cutaneous melanoma. Br J Dermatol. 2002;147:62–70. 41. Garbe C, Buttner P, Bertz J, et al. Primary cutaneous melanoma. Identification of prognostic groups and estimation of individual prognosis for 5093 patients. Cancer. 1995;75:2484 –2491.