In the 8th edition of the textbook Ackerman's Surgical Pathology edited by Juan Rosai,⁴² and published in 1989, general guidelines are offered for handling tissue received from a dissection of nodes undertaken in search of cancer. First, a pathologist is advised to examine grossly the entire specimen and attempt to identify and harvest all nodes in the tissue submitted (nodes less than 3 mm in greatest diameter are submitted in their entirety, whereas nodes larger than that are divided into slices of 2–3 mm each; a representative slice sufficiently large to fit neatly in a single cassette is processed). The nodes then are processed and sections of tissue cut from them are stained with hematoxylin and eosin (H&E), after which they are studied by conventional microscopy. Prior to the advent of SNB, the manner advised by Rosai was the one by which nodes that might contain a metastasis were handled routinely. It was understood that examination of a particular node by microscopy in this "hit or miss" fashion might result in failure to detect a metastasis of melanoma. That very limitation was demonstrated aptly in a study conducted by Cochran and associates⁴³ of 2,227 nodes removed from 100 patients who had undergone END for melanoma. Scrutiny of sections of tissue stained with H&E revealed no evidence of melanoma in those nodes. When the tissue then was stained with antiserum to S100 protein, which heightens greatly capability for detection of cells of melanoma by conventional microscopy, 16 nodes from 14 patients were found to contain melanoma. This reality was confirmed by use of NK1/C3, a monoclonal antibody that attaches to cells of melanoma. The authors acknowledged that the significance clinically of cells of melanoma being detected in nodes by use only of immunohistochemical (immunoperoxidase) stains had yet to be realized. Nonetheless, they made this telling comment: " . . . of great interest [is] that the 14 patients in this study who had OMC [occult cells of melanoma] had a death rate twice that of patients without OMC."

The advent of SNB prompted observations with intensity to be made of that node said to be the sentinel one. The sentinel node, according to Cabanas,⁴⁴ is the node to which a malignant neoplasm first metastasizes, whereas the sentinel node in the opinion of Morton and coworkers⁴⁵ is "the lymph node nearest the site of the primary melanoma, on the direct drainage pathway." In the first study of the subject of SNB in patients with melanoma by Morton, et al.⁴⁵ in 1992, all nodal tissue harvested from both the sentinel nodes (SNs) and the lymphadenectomy specimen was examined by conventional microscopy. The SNs were bisected through the hilum, sections from one half of the specimen being submitted for examination by frozen techniques and sections from the other half being processed for staining rapidly with immunoperoxidase. All of the SNs and the secondary nodes, that is, nodes not considered to be sentinel ones but occurring in the same nodal basin, were fixed in formalin, stained with H&E, immunoperoxidase for S100 protein, NK1/C3, and HMB45, and examined by conventional microscopy. Forty SNs were found to house cells of melanoma. Seventeen of the 40 metastases were detected by immunoperoxidase stain alone.

Examination of the SN by frozen section, which was the method advanced by Morton and coworkers, afforded an opportunity to extirpate the nodal basin if, at the time of SNB, the SN harbored neoplastic cells. From the outset, frozen section was the standard manner in which the SN was examined initially. An alternative was to perform SNB, process the SN in routine fashion, and, if the SN was found to contain cells of melanoma, remove the nodal basin in a second operation.

Gibbs, et al.⁴⁶, in 1999, shared their experience with 69 consecutive patients with melanoma who underwent SNB, a frozen section having been performed on the SN of 64 of them. Fourteen of those nodes were found by standard techniques to harbor cells of melanoma. Of those 14, the correct diagnosis of metastatic melanoma was made by frozen section in four instances, an equivocal diagnosis was rendered in two instances, and no cells of melanoma were observed in sections from the remaining eight specimens. When sections of tissue from the eight nodes prepared by frozen section technique were reviewed, three of them were found to be positive for melanoma, an indication that they had been misread histopathologically at the time of operation. The other five nodes were negative because no neoplastic cells were present in frozen sections of tissue, but they were detectable after processing of tissue that had been fixed in formalin. The authors concluded that "Intraoperative FS [frozen section] has low sensitivity in identifying MM [melanoma] within the SLN [sentinel lymph node]."

The efficacy of frozen section technique for identifying cells of melanoma in a node removed by SNB was assessed by Tanis, et al.⁴⁷ who, in 2001, examined 177 SNs from 99 patients with melanoma and 444 nodes from 262 patients with carcinoma of the breast. The results obtained from examination of frozen sections were compared with those derived after routine processing and staining of tissue by H&E and immunohistochemical methods. Examination of frozen sections permitted detection of melanoma in only 8 of 17 nodes that, in fact, housed a metastasis of melanoma. In like manner, carcinoma was detected in only 71 of 96 nodes that actually contained a metastasis of carcinoma of the breast. The sensitivity of examination by frozen section was 47% in patients with melanoma and 74% in patients with carcinoma of the breast. The authors opined that "Frozen section analysis is not recommended in patients with melanoma."

Results similar to those of Gibbs, et al. and Tanis, et al. were obtained by Koopal, et al.⁴⁸ Metastasis of melanoma was discovered in frozen sections in five of 17 SNs that contained melanoma (a sensitivity of 38%), constraining the authors to recommend the following: "The combination of the low sensitivity of FSA [frozen section analysis] and a finding that only 12% of the SLNBs [sentinel lymph node biopsies] contained metastases does not justify routine use of FSA on the sentinel nodes of patients with cutaneous malignant melanoma."

The reasons for discrepancies apparent in these studies are explicable. The technique that utilizes frozen section is much less refined than the one employed for processing tissue routinely after it has been fixed in formalin. It is difficult technically to obtain adequate sections of tissue for interpretation after that tissue has been frozen. Artefacts consequent to freezing tissue and other technical limitations, such as sections that are thick and stained badly, may hamper "reading" of sections accurately. In an attempt to obviate those limitations, tissue often is trimmed to such an extent that considerable wastage of it occurs. The tissue squandered may be the only material that really contains the metastasis of melanoma and, as a result, an opportunity for coming to an accurate diagnosis may be lost by virtue of tissue that has perished. As a rule, only a single section is examined when "frozen technique" is performed, in contrast to numerous sections when "routine technique" is employed; that limitation, too, decreases the chance of detecting cells of melanoma. As these various constraints became evident increasingly, frozen section gradually lost favor as a technique for assessment of SNs and, eventually, was abandoned. To complicate matters further, if a SN contained radioactive tracer, the pathologist and other laboratory staff were at risk for exposure to hazards of radiation. This threat required development of complicated protocols designed to deal with a SN during frozen section, which added to the ever-increasing unpopularity of frozen section technique for examination of nodes.

An "imprint slide," that is, a glass slide to which cells have adhered consequent to pressing the slide against tissue with transfer thereby of cells to the slide, may be used to assess a SN at the time of surgery. This technique avoids wastage of tissue at the same time that it does not compromise detection of cells of melanoma. But the imprint technique permits only a small number of cells to be examined, is insensitive, and is subject to false negative results. Those drawbacks make imprint technique undesirable unless the SN appears to contain a metastasis on examination grossly and unless there is a need for immediate determination preliminarily of the status of the node. In that regard, Messina, et al.⁴⁹ commented in 1999 as follows: "Intraoperative touch preparation cytology may be used as an adjunct technique in sentinel nodes grossly suspicious for metastatic disease. This technique has been performed on 23 sentinel nodes, with no false positives and an overall sensitivity of 62%." If a SN contains radioactive tracer, concerns arise about the hazards of exposure to radiation and, for that reason, intraoperative cytopathological examination seldom is used in the assessment of a SN.

The limitations of the techniques of frozen section and imprint examination have led to virtual elimination of intra-operative consultations in the initial assessment histopathologically of a SN. In brief, those limitations are: (1) Decreased sensitivity because tissue is represented inadequately, (2) errors in sampling, (3) wastage of tissue, and (4) exposure potentially to radiation. Those two aforementioned technologies still may be employed if, for example, a surgeon believes that after having examined a SN physically, it is likely to contain neoplastic cells. A frozen section then may be requested for the purpose of confirming or denying that impression prior to extirpation by the surgeon of the nodal basin.

The best method available currently for examination of a SN is processing of tissue in routine fashion, fixing the tissue in formalin, embedding it in a block of paraffin, cutting sections with a blade of a microtome, and placing those sections on a glass slide prior to staining them with H&E and examination of them by conventional microscopy. Unlike the situation in dissection of regional nodes, the entire SN is cut through by a technician in an effort to permit cells of melanoma to be detected, no matter how few of them there may be. Because not every bit of tissue cut is placed on slides, a sliver of tissue, unexamined, still may contain cells of melanoma. In an effort to spy whatever neoplastic cells may exist in a sentinel node, processing in routine fashion and staining by H&E have been complemented by new technological advances, among them those of immunology, molecular biology, and cell culture.

Immunohistochemistry utilizes antibodies to bind to antigens that may be present on a neoplastic cell. Attached to those antibodies are an "indicator" that conveys to an observer whether an antibody-antigen reaction has taken place. The technique demonstrates antigens on cells by decorating them with a brown or red dye that is observable readily by conventional microscopy, thereby lending itself to identifying neoplastic cells with specificity; in the case of the SN, this technique facilitates detection of very small numbers of cells of melanoma (Figure 5). The antigens employed most commonly for the purpose of such detection are S100, HMB45, Melan A (Mart 1), and NKI/C3. Because those antigens are neither 100% specific nor sensitive, and because of the need to circumvent that limitation, the antigens often are used together in a "cocktail." The technology itself may be "temperamental," making it difficult at times to produce sections of tissue of excellent quality, which is yet another limitation of the technique pertinent to SN. S100 protein is expressed on normal cells, such as melanocytes, nerves, cartilage, and fat, as well as on cells of neoplastic counterparts of them. In lymph nodes, S100 protein is expressed in cells found there normally, namely, interdigitating and dendritic reticulum cells and even nerves, as well in cells that may appear there unexpectedly, such as abnormal melanocytes of a nevus and abnormal melanocytes of melanoma. HMB45 is of value only for cells of melanoma, and Melan A only for abnormal melanocytes of both benign and malignant neoplasms. In the case of SN, it is crucial to be able to differentiate abnormal melanocytes of a nevus from abnormal melanocytes of a melanoma. That is accomplished by being alert to the precise location of the cells within the node, the expression of antigen on those cells, the silhouette of aggregations of cells, and the cytomorphologic attributes of them. Melanocytes of melanoma are situated in the subcapsular sinuses near the center of the node, stain variably with antigens for melanoma, are arranged in aggregations that vary in size and shape and often assume bizarre geometric outlines, and nuclei of them tend to be crowded and pleomorphic. Abnormal melanocytes of a nevus usually are lodged within the capsule of a node, are S100 positive, may be melan A positive, rarely are HMB45 positive, are organized in discrete nests that are relatively uniform in size and shape, and nuclei of them are monomorphic. Reticulum cells are found in paracortical zones or in germinal centers and are positive for S100 protein but negative for HMB45, NKI/C3, and melan A. Nerves are S100 positive, may be melan A positive, and rarely are HMB45 positive. As a rule, immunohistochemical stains are performed only if stain with H&E fails to reveal cells of melanoma in the node. If immunohistochemical stains indicate that cells positive for antigens of melanocytes are present, sections of tissue immediately adjacent are then stained with H&E, and usually will be shown to contain cells of melanoma. Determination of whether those cells truly are abnormal melanocytes of a melanoma or abnormal melanocytes of a nevus, or are dendritic cells, can then be made.

Cochran, et al.,⁴³ in their study published in 1988 prior to the era of SNB, used immunohistochemical stains for nodes that were negative for metastatic melanoma as judged by examination of sections stained with H&E. Using those specialized stains, they were able to detect a metastasis of melanoma in 16 of the nodes. On the basis of the experience of Cochran and coworkers, immunohistochemistry is now used routinely in the assessment of a SN for the purpose of detecting foci of melanoma in it. Gershenwald⁵⁰ reviewed what were thought initially to be a negative SN in 27 patients who subsequently developed overt metastases of melanoma. When the examination was repeated and the nodes sectioned serially and stained by immunohistochemical methods, melanoma was diagnosed in 16 of 27 patients. Eight melanomas were discovered in sections stained by H&E and eight were found by immunohistochemical stains. Gadd⁵¹ and Clary⁵² each identified seven patients with metastases of melanoma whose SN was sectioned serially and stained immunohistochemically, thereby permitting a small number of cells of metastatic melanoma to be discerned in three patients and in four patients, respectively. It is apparent that the more sections of a SN examined by H&E and immunohistochemical stains, the greater is the likelihood that cells of melanoma will be detected in them. Sectioning through an entire SN is time consuming and, for that reason, is impractical. Cochran⁵³ asserted that if after the center of a SN is sampled adequately and examined carefully, and sections of it are not found to contain cells of melanoma, it is unlikely that cells of metastatic melanoma will be present elsewhere in the node. He advised further that in his experience a metastasis of melanoma grows first in the center of the node and then spreads to the periphery of that node. That being the case, according to him, scrupulous examination of the central zone in search of cells of melanoma usually will reveal those cells if they are truly present. As pathologists became progressively more experienced⁵³ with examination of SNs, the rate of detection of metastases increased in sections stained by H&E, therefore lessening the need for stains that utilize immunohistochemistry. A protocol proposed by Cochran^53,54 for handling SNs is based on the experience of his group with 1,119 of those nodes. The coworkers no longer use frozen sections routinely; the SN is bisected and each half is processed in the very same manner. This is how it is done by them: 10 serial sections are readied and sections 1,3,5, and 10 are stained with H&E; section 2 is stained for S100 protein and section 4 for HMB45 or Melan A; sections 6 and 7 are negative controls for immunohistochemical stains, and sections 8 and 9 are retained as "spares" to be used, if necessary, to repeat studies that technically are unsatisfactory. Immunohistochemical stains are performed only if stains for H&E are negative for cells of melanoma. Staining with H&E and with immunohistochemical stains is now routine for detection of "microscopic" or "occult" neoplastic cells in sections of biopsy specimens of SNs.

Cell culture and molecular biologic techniques (Figure 6) are emerging as other ways of detecting a metastasis of melanoma when examination of a node by conventional microscopy is negative. At the end of the twentieth century, molecular biologic studies were introduced in order to spot cells of melanoma in nodes, peripheral blood, and bone marrow of patients who had a melanoma in the skin. These methods seemed to be more sensitive than were conventional techniques for detecting cells of melanoma, and they appeared to have significance clinically.

Heller, et al.,⁵⁵ in 1991, wrote about the use of cell culture to detect neoplastic cells in nodes of patients with melanoma. They examined 323 nodes taken from 41 patients. Cell culture enabled detection of cells of melanoma in 155 of those nodes. Assessment of the nodes in routine fashion by conventional microscopy revealed cells of melanoma in 20 of 323 nodes; 18 of the nodes were positive, too, by cell culture. No melanoma was detected using conventional microscopy in nodes of 29 of 41 patients. Cell culture, however, detected melanoma in nine patients whose nodes did not contain neoplastic cells by conventional microscopy. In three of the nine, metastases became apparent after a follow-up period whose mean time was 18 months. All patients in whom nodes were negative by cell culture and by examination by conventional microscopy seemed to be free of metastases after 18 months. The results of this study, like those that utilized molecular technologies, suggest that even when cells of melanoma are not detected by methods that employ conventional microscopy, including immunohistochemical technique, they may be present, nonetheless, in nodes.

In 1998, Schrivers, et al.⁵⁶ reported on their results in assessing SNs using the Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR). In 23 of 114 patients, cells of melanoma were observed in SNs both by conventional methods and by RT- PCR techniques. In 47 other patients, RT-PCR was able to detect cells of melanoma even though no evidence of them had been found in SNs by routine examination histopathologically. In six of those patients, metastases appeared elsewhere subsequently. The remainder of the patients had no evidence of melanoma in their SN, either by conventional microscopy or by the RT-PCR; in one of those patients, metastases became manifest later. This is what the co-authors had to say about their experience: "The current study shows that the tyrosinase RT-PCR assay is more sensitive than routine histological examinations for the detection of SLN micrometastases . . . This article shows a statistically significant difference in both disease free (P=.02) and overall (P=.02) survival between patients with histologically negative results with a PCR-negative SLN and those with a PCR-positive SLN."

Bostick, et al.,⁵⁷ published in 1999, their findings in sections stained by H&E and by immunohistochemical stains of specimens of SNs removed from 72 consecutive patients with melanoma. In addition, they utilized RT-PCR for that same purpose. The polymerase chain reaction employed three mRNA markers, namely, MAGE-3, Mart-1, and tyrosinase mRNA, for identifying cells of melanoma in the SN. Seventeen of the patients had a metastasis in their SN, diagnosed either by H&E or by immunohistochemical stains. Sixteen of the 17 SNs were positive for at least 2 mRNA markers. Twenty (36%) of the patients whose SNs were negative both by H&E and immunohistochemical stains were positive for at least two mRNA markers. Three of those 20 (15%) developed manifestations of metastatic melanoma clinically. No patient whose SN was negative histopathologically and that expressed one or no mRNA markers developed a metastasis thereafter. The single patient who was said to have cells of melanoma in a SN, but in whom markers of mRNA could not be demonstrated, did not develop any sign of a new metastasis. In this series the period of follow up was 12 months, which is insufficient to determine the significance of the positive mRNA markers in the 17 other patients who had negative SNs as judged by conventional microscopy. Nevertheless, it can be concluded, as Bostick and co-workers did, that positive RT-PCR markers do have significance clinically. This is what they said in that regard: "MM [multiple- RNA markers] expression in the SN more accurately reflects melanoma micrometastases and is also a more powerful predictor of disease relapse than are H&E staining and IHC [immunohistochemistry] alone."

In 1999, Lukowsky, et al.,⁵⁸ using conventional microscopy and tyrosine RT-PCR, examined the SN of patients with melanoma. Ten patients had cells of melanoma in the SN as determined by conventional microscopy. RT-PCR revealed that three other patients also had a SN positive for melanoma. Eleven of those 13 patients had positive secondary nodes as indicated by findings in one or both of the tests. Seven patients had a SN that was negative both by conventional microscopy and by RT-PCR. Six of those patients had positive secondary nodes by PCR only; one had positive secondary nodes both by conventional microscopy and RT-PCR. The authors interpreted the results in this way: "Thus, up to 30% of patients may harbor metastasis despite the non detection of melanoma cells in the SLN [sentinel node]." They also issued a recommendation thus: "Although the number of melanoma patients investigated in our study is too small yet for a clinically relevant conclusion, a further evaluation of the SLN concept is recommended by our findings."

Studies of SNs in archival material, that is, tissue embedded in paraffin, and the peripheral blood of 75 patients were reported on in 2001 by Palmieri, et al.⁵⁹ using RT-PCR. One of the stated purposes of the study was to assess the ability of RT-PCR to detect cells of melanoma in a SN that already had been processed and found to be negative histopathologically for those cells and in the peripheral blood of those patients, as well as of patients with a positive SN. The other was to determine if positivity by RT-PCR for cells of a melanoma in a SN had prognostic value. In that endeavor, they employed tyrosinase and Melan A as markers for melanoma and examined 92 SNs obtained from 75 patients with melanoma. Sixty-one patients had no sign of cells of melanoma in a SN as evidenced by immunohistochemical stains, whereas 14 patients did have melanoma in a SN. Thirty of 61 SNs that did not contain cells of melanoma when assessed histopathologically were positive for at least one mRNA marker. An unexpected finding was that four of the 14 SNs that contained melanoma on examination by conventional methods were negative for mRNA markers. Although the authors suggested that this may have been the result of degradation of the markers in the tissue embedded in paraffin, that is mere speculation. The peripheral blood of 20 patients expressed the tyrosinase marker and this seemed to correlate with the stage of the disease clinically and the risk for developing additional metastases overtly. The presence of the RNA markers correlated with a significant increase in risk for appearance of metastases. The authors characterized that risk in these words: " . . . poorer DFS [disease free survival] and OS [overall survival] were registered in patients with histopathologically negative SLNs that expressed both mRNA markers." And this was their conclusion: "RT-PCR analysis of serial sections from archival SNs could support conventional histopathological methodologies (largely demonstrated to underestimate the real incidence of malignant melanoma metastases) in improving detection of occult metastases."

Blaheta et al.,⁶⁰ in 2000, told of having examined sections of SNs of 116 patients using H&E and immunohistochemical stains for HMB45 and tyrosine RT-PCR. Thirteen percent (15) of patients had a positive SN by conventional microscopy and by RT-PCR; 31 percent (36) had a positive SN for PCR alone and the remainder (65) had a SN that was negative for cells of melanoma by all methods employed. After a period of follow up of 19 months, 67% (10) of patients who had a positive SN by both conventional histology and RT-PCR, 25% (9) of patients who had a positive SN by RT-PCR only, and 6% (4) of patients with a negative SN developed obvious metastases of melanoma clinically. The results of this study seem to indicate that detection by PCR techniques of cells of melanoma in a SN is synonymous with metastasis of melanoma beyond the SN, even in the absence of any evidence of metastasis by conventional microscopy.

Markers for melanoma using the RT-PCR technique also may demonstrate cells of melanoma in bone marrow and in peripheral blood, even when the SN is negative by both routine examination histopathologically and by examination using RT-PCR. Blaheta, et al.,⁶¹ in 2001, reported on their examination of bone marrow and peripheral blood of 26 patients with melanoma in whom a SNB had been performed. Thirty-five SNs, 41 samples of bone marrow, and 26 specimens of peripheral blood were examined by RT-PCR for tyrosinase and Melan A. The SN and bone marrow were studied, in addition, by conventional microscopy. Scrutiny of sections by conventional microscopy revealed cells of melanoma in the SN of four patients. Those four patients and three others had a SN that was positive for melanoma by RT-PCR. Melanoma was detected by RT-PCR in the bone marrow of two patients and in the peripheral blood of six patients. Nothing was mentioned about survival of those patients, but the positivity by RT-PCR for cells of melanoma in a SN and in bone marrow seemed to correlate with the thickness of the primary melanoma, suggesting that a positive RT-PCR at those anatomic sites may be associated with a poor outcome. A positive RT-PCR in the peripheral blood was not related to the thickness of the primary melanoma.

Exhaustive sectioning and scrupulous examination of SNs have led to the conviction that nodes negative by conventional microscopy but positive by RT-PCR may not be a consequence merely of sampling. Blaheta, et al.,⁶² in their study reported on in 2001, re-examined 33 SNs from 21 patients, all of those nodes showing positivity for tyrosinase RT-PCR, but at the time when first examined showed no evidence of melanoma by conventional microscopy. The re-examination, which included deeper sections stained with H&E or immunohistochemical stains, revealed cells of melanoma in only a single SN. Six of the remaining 20 patients developed metastases of melanoma during a median follow-up time of 34 months.

The studies by Schrivers, et al.,⁵⁶ Bostick, et al.,⁵⁷ Palmieri, et al.,⁵⁹ and Blaheta, et al.,^60,62 demonstrate that detection of cells of melanoma by molecular techniques only are significant because some patients whose SNs were positive for melanoma by RT-PCR, yet did not seem to contain melanoma by conventional microscopy, eventually developed overt metastases and died from the effects of them. Those patients harbored in their SN cells of melanoma that evaded detection by routine examination using conventional microscopy, that is, using sections stained with H&E and immunohistochemical stains. Because some of the cellular products detected by the molecular techniques are not specific for cells of melanoma and may be present on abnormal melanocytes of a nevus, a positive reaction does not always indicate metastasis of melanoma.

Our search of the literature failed to uncover any article in which the sensitivity of cell culture for detecting metastatic melanoma was compared to the sensitivity of molecular techniques employed for that purpose. Cell culture has not been applied to the investigation of SNs in patients with melanoma. The study of Heller and coworkers⁵⁵ has not been validated and results of attempts to reproduce it have not been published. Cell culture is expensive, labor intensive, and slow, and those may be the main reasons why it has not been brought to bear seriously on the matter of SNs.

In short, the detection of cells of melanoma in nodes, especially SNs, has evolved from employment of conventional techniques of tissue fixed in formalin, processed in routine fashion, and then stained with H&E and with immunohistochemical methods, to utilization of more advanced techniques predicated on principles of molecular biology. Frozen section and imprint cytopathology seldom are used nowadays. It has become obvious that the more intense and intensive the search for cells of melanoma in a SN the greater is the likelihood of discovery of them. Markers for cells of melanoma in nodes, peripheral blood, bone marrow, and other organs enable identification of melanoma there and, perforce, have established beyond doubt the reality of primary melanoma in the skin having metastasized, even if the metastases have not yet manifested themselves clinically.