Thin Melanoma. N Context. The incidence of malignant melanoma is

Thin Melanoma David E. Elder, MB ChB, FRCPA N Context. The incidence of malignant melanoma is increasing and a preponderance of the melanomas diagnosed today are thin in terms of Breslow criteria. Although thin melanomas, as a group, are associated with a very good prognosis, a subset of these tumors may metastasize and cause death. These cases can be identified by using prognostic models, including the standard American Joint Committee on Cancer criteria, and other attributes identified in follow-up studies. Objective. To review the history of concepts of prognostic modeling in melanoma, focusing on thin melanomas. Data Sources. Selected literature. Conclusions. About 40 years ago, it was realized that malignant melanoma, once almost uniformly fatal, could be divided into categories with better or worse prognosis through the use of prognostic models. The first simple models, Clark levels of invasion and Breslow thickness, are still in use. Thickness remains the single most useful variable. Breslow recognized that melanomas less than 0.76 mm in thickness were associated with a very good prognosis, with no metastases in his limited initial study. The American Joint Committee on Cancer selected a cutoff of 1.0 mm, which achieves a similar result, with stage modifiers, although some metastases and deaths do occur with stage I lesions. Clark demonstrated an almost equally good prognosis for his level II invasive melanomas and recognized that most of these lesions, although invasive, lacked the ability to form tumors or to undergo mitosis in the dermis and were therefore nontumorigenic and nonmitogenic and lacked competence for metastasis. Studies of these low-risk melanomas have led to the development of criteria for earlier diagnosis and a steady, but still inadequate, improvement in prognosis for melanoma overall. Multivariable models currently can identify groups of patients within the thin melanoma category whose prognosis varies, from a disease-free survival of close to 100% to about 70%. Prognosis declines more or less linearly with increasing thickness, modified by ulceration, mitotic rate, and other attributes. (Arch Pathol Lab Med. 2011;135:342 346) LEVELS OF INVASION AND THICKNESS AS PROGNOSTIC MARKERS IN MELANOMA Although for most tumors, diameter is a powerful prognostic attribute, this is not so for melanomas, which can be very broad and yet have a good prognosis. This is because most melanomas have an extensive in situ or superficially invasive component that does not contribute to metastatic potential. It is therefore misleading and not recommended to report the breadth of a melanoma in a pathology diagnosis. The extent of penetration of melanoma cells into the skin was recognized as a better predictor of survival than breadth in a few early studies 1,2 and was first codified into a powerful model in 1967 by Clark, who described a system of 5 levels of invasion: level I, in which the lesion is confined to the epidermis; level II, in which the neoplasm extends into the papillary dermis (which is a stroma that is specialized to support epithelium) either as a few scattered cells or small nests; Accepted for publication March 4, 2010. From the Department of Pathology and Laboratory Medicine, Division of Anatomic Pathology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania. The author has no relevant financial interest in the products or companies described in this article. Reprints: David E. Elder, MB ChB, FRCPA, Department of Pathology and Laboratory Medicine, Division of Anatomic Pathology F6060 HUP, University of Pennsylvania School of Medicine, 3400 Spruce St, Philadelphia, PA 19104 (e-mail: elder@mail.med.upenn.edu). level III, in which a tumor is formed that fills and expands the papillary dermis; level IV, in which the tumor invades the reticular dermis; and level V, in which the subcutis is invaded. The survival of patients with melanomas of Clark level II was much better than that for patients with tumors at deeper levels 3,4 ; although the tumor had metastasized for 3 of 36 patients with level II melanomas in the initial study, subsequent studies showed a much superior survival rate of approximately 98% when polypoid melanomas are excluded. In 1970, Breslow described a simple system for measuring the thickness of melanomas. 5 The Breslow thickness is determined by measuring from the top of the granular layer of the overlying skin, or from the base of an overlying ulcer, to the deepest invasive melanoma cell. In Breslow s study of 98 patients, the survival of patients with thin melanoma, which he defined as a melanoma less than 0.76 mm in thickness, was 100%. Of 45 patients with Clark level II melanomas in Breslow s study, only 1 had a tumor that metastasized, resulting in a disease-free survival of 98%. In studying tumor thickness, Breslow was initially motivated by attempting to combine this measurement with that of the breadth of the lesion as an estimate of tumor volume. However, it soon became apparent that thickness was most valuable as a single variable. These studies by Clark and Breslow resulted in a paradigm shift in the management of melanoma. Before, it was generally assumed that a diagnosis of melanoma was 342 Arch Pathol Lab Med Vol 135, March 2011 Thin Melanoma Elder

a death sentence. The recognition of the existence of a readily recognizable subset of melanomas associated with a good prognosis made it possible to contemplate efforts at prevention of mortality by diagnosis of melanoma in its early, curable stage. 6 In addition, the therapeutic approach to melanoma was modified to take into account the new information, resulting in a marked reduction in the extent of surgery that was recommended for the disease. 7 11 PHASES OF TUMOR PROGRESSION: RADIAL AND VERTICAL GROWTH PHASE In the earliest descriptions of melanoma, the disease was regarded as a tumor mass that developed in the skin and was often surrounded by satellites at the time of presentation. Mortality was very high, and therapy was very aggressive, yet ineffective in preventing mortality. A few studies recognized that more superficial melanomas have a better prognosis. One study, 1 for example, distinguished melanomas that have a flare of pigmentation from those that did not. This pattern of superficial spreading growth was recognized the most clearly by Clark, who referred to it as the radial growth phase (RGP) of melanoma development. 3,4 This term was said by Clark to have been first proposed by Scott Blois, MD, at a meeting in San Francisco. It was intended to refer to a pattern of spread of a patch or plaquelike lesion, gradually expanding along the radii of an imperfect circle in the skin. The term radial growth phase is clinically derived. To a histopathologist, the RGP appears to be more linear than circular. The tumorigenic vertical growth phase (VGP), when present, forms a mass that may histologically assume a more or less circular shape and may thus appear radial. Therefore, some histopathologists have referred to the RGP as the horizontal growth phase. In our practice, however, we continue to use the originally defined term radial growth phase. Because a tumor mass is not formed in the RGP, its prognosis is excellent. The observed survival in a study of 65 invasive melanomas that lacked VGP 12 was literally 100%; subsequent studies 13 have shown that 1% or less of tumors may metastasize, and in our experience all such lesions have had extensive regression that may have down-staged a preexisting tumorigenic VGP component. The term vertical growth phase was proposed at the same time as RGP to refer to a qualitatively different phase of tumor growth in which the development of a property of expansile rather than spreading growth led to the formation of a tumor in the true sense of the word, namely a swelling or mass. It was emphasized by Clark that this did not represent a gradual transition of the preexisting RGP, but represented the development of a qualitatively new pattern of tumor growth, which was visible clinically and pathologically within the preexisting RGP in many cases. In contrast to the cells of the RGP, those of the VGP have acquired the ability not only to survive but also to proliferate in the dermis and are tumorigenic. In a minimal lesion, the melanoma is classified as tumorigenic if a cluster or nest of cells in the dermis is larger than the largest nest in the epidermis, in terms of the number of cells across its greatest diameter. This is consistent with the presumption that proliferation of cells in the nest has occurred in the dermis, after migration of the nest from the epidermis. When the sizes of the nests are equal, we term the lesion probable early tumorigenic vertical growth phase, to keep the RGP category with its excellent prognosis as pure as possible (Figures 1 and 2). A study using the Ki-67 proliferation marker 14 demonstrated, as predicted by the model defined above, that proliferation is confined to the epidermis in the RGP and is present in both epidermal and dermal compartments in the VGP. A prognostic model for VGP cases demonstrated that the single strongest independent predictor of survival was the mitotic rate expressed in mitoses per square millimeter. 15 The property of mitogenicity, or the presence of any mitoses in the dermis, was included in the definition of VGP because a few cases were found to have metastasized that were mitogenic but not tumorigenic. 16,17 The Clark levels of invasion and phases of progression are closely related but not synonymous. It is possible for an early vertical growth phase to be tumorigenic and to expand the papillary dermis, but not to fill it, qualifying as level II invasion (Figures 1 and 2, a and b). In some cases, the expansion may be considerable, forming a polypoid mass. These lesions were termed Clark level III, by convention. The vertical growth phase is often elevated above the skin. The depth of penetration into the dermis is variable, but the extent of invasion below the original surface level of the skin is often limited and many lesions are confined to the papillary dermis and/or the superficial reticular dermis. Thus, much of the thickness of the tumor is derived from cells that are above the original level of the skin surface (Figure 3). This observation may explain why thickness, which measures the entire extent of the vertical growth phase (up as well as down), has been found to be a superior measure to the level of penetration into the skin for predicting survival. However, in thin melanomas, thickness does not necessarily discriminate between radial and vertical growth phase. PROGNOSTIC MODELS FOR THIN MELANOMA Based on these considerations, it is clearly of value to recognize melanoma when in its early, curable stage. This stage of curability can be recognized in several different ways, including the use of Clark criteria for level I or level II, the criteria for nontumorigenic RGP, Breslow thickness criteria, and combinations of these features. Studies over time have shown tumor thickness to be the simplest and the most reliable as a single variable. Therefore, current American Joint Committee on Cancer (AJCC) staging systems use Breslow thickness as the primary staging attribute, with the use of stage modifiers, of which ulceration was the most important in the recently superseded 2001 system. 18 However, in this and in the newly released present system, a cutoff of 1 mm rather than the Breslow original cutoff of 0.76 mm was used. As a result, the AJCC stage IA tumors, those with the best prognosis, nevertheless had a 10-year mortality rate of 12% in the 2001 AJCC model, although in a more recent population-based study 19 the mortality was about half this rate. The properties of tumorigenicity and mitogenicity, as well as Clark levels of invasion, and regression, have been evaluated as modifiers of thickness for prediction of survival in AJCC thin melanoma. In a 2004 study of 884 patients with thin melanomas, Gimotty et al 13 demonstrated that both tumorigenicity and mitogenicity (mitotic rate.1) were independently predictive of survival. A model was developed that was superior by formal statistical analysis to the AJCC staging model, in terms Arch Pathol Lab Med Vol 135, March 2011 Thin Melanoma Elder 343

Figure 1. A superficial spreading melanoma characterized by nested and pagetoid proliferation of uniformly atypical epithelioid melanocytes in the epidermis. A few clusters of similar cells are present in the papillary dermis, which is neither filled nor expanded. None of the clusters (nests) in the dermis is larger than the largest nest in the epidermis, and there are no dermal mitoses, qualifying this lesion as nontumorigenic, nonmitogenic radial growth phase only, with an excellent prognosis (hematoxylin-eosin, original magnification 310). Figure 2. a, A broad lesion characterized by extensive proliferation of large epithelioid melanocytes in the epidermis, with extensive pagetoid scatter into the epidermis, and by a few nests of similar cells in the papillary dermis, which is expanded but not filled, corresponding to Clark level II invasion. b, The largest of the nests is about the same size as the largest nest in the epidermis, qualifying it as probable early vertical growth phase. Nevertheless, this nonulcerated, nonmitogenic, Breslow thin lesion (0.37 mm) from the back of a 47-year-old woman is associated with an excellent prognosis (see the Table) (hematoxylin-eosin, original magnifications 32.5 [a] and 35 [b]). Figure 3. In this image of a thick polypoid melanoma, the bulk of the tumor is above the original level of the skin, filling and expanding the papillary dermis, and qualifying as Clark level III and as tumorigenic and mitogenic vertical growth phase. The tumor has presumably enlarged by a process of addition of younger cells at the surface. The Breslow thickness measurement captures the full extent of the lesion by measuring up as well as down (hematoxylin-eosin, original magnification 31). of predicting survival. Tumor mitotic rate was a powerful attribute in both a 1989 Clark model based on only VGP cases15 and a 2005 Gimotty model based on thin melanomas.14 Mitotic rate as a prognostic attribute was first introduced into the literature by Cochran20 in 1967 and was also studied by Schmoeckel and Braun-Falco,21 who proposed a mitotic index based on the product of mitotic rate and thickness. More recently, a large study of mitotic rate from the Sydney Melanoma Unit22 also supported its usefulness. This group also demonstrated the reproducibility of mitotic counting.22,23 As a result of these studies, the working group for the revision of the AJCC staging system has conducted its own study of mitotic rate and has found this to be an important discriminator of survival, especially in thin melanomas. The group decided to add the criterion of mitotic rate greater than or equal to one (or mitogenicity) as a modifier of AJCC stage I in the current AJCC staging system.24 It should now therefore be routine for pathologists to record this important attribute. The guidelines suggest that mitotic rate should be determined by first examining the tumor and identifying a hot spot of 344 Arch Pathol Lab Med Vol 135, March 2011 greatest mitotic activity if one exists. Then, the number of mitoses should be counted in contiguous high-power fields totaling 1 mm2. In most microscopes, this will be about 3 to 4 high-power fields; however, the field area should be determined in individual microscopes by determining the diameter of a high-power field by using a stage micrometer or a ruler and with the formula pr2. The rate should be given as a whole number. In lesions in which the total amount of tumor in the dermis is less than 1 mm2 (or even less than 1 high-power field), the number of lesional cell mitoses in a square millimeter of dermal tissue containing the tumor should be determined. Stated differently, the number of mitoses present in the dermal tumor should be reported. Again, this will always be a whole number. For 2 reasons, it is not recommended that multiple step sections be prepared to absolutely rule out the presence of mitotic activity. First, the studies that demonstrated the importance of mitotic rate were based, for the most part, on melanomas diagnosed routinely in the community, and then referred to large centers where the prognostic studies were done. Second, in the coming age of targeted therapy, it is desirable to preserve tissue in Thin Melanoma Elder

Expanded Classification for Thin Melanomas (, 1 mm) a,b Category SEER Database (10-y Survival Rate) No. of Cases Survival, % Not ulcerated Level II, thickness,0.78 mm, age,60 y 10 648 99.0 Level II, thickness,0.78 mm, age.60 y 5258 97.5 Level III, thickness,0.78 mm, other sites 4169 96.8 Level III, thickness,0.78 mm, head and 664 92.1 neck Level II/III, thickness.0.78 mm, women 1397 95.6 Level II/III, thickness.0.78 mm, men 1608 90.6 Level IV/V 2213 91.4 Ulcerated Level II/III 215 88.9 Level IV/V 119 69.8 a A classification of thin (American Joint Committee on Cancer stage I) melanomas by using Breslow thickness, Clark levels if invasion, ulceration, sex, and lesion location as attributes. b Data from US population-based SEER tumor registry. 25 Abbreviation: SEER; Surveillance, Epidemiology, and End Results. the block in case genotyping is needed if metastases should develop in the future. At present, most databases do not include mitotic rate as an attribute. Gimotty and colleagues 25 therefore developed a model for thin melanoma of unknown mitogenicity that discriminates a very large subset of patients for whom metastasis is very rare indeed, and a small subset for whom the risk of metastasis is quite substantial. The former group can therefore be spared aggressive therapy, while the latter group may be offered adjuvant therapy and may be considered for sentinel lymph node sampling. In developing this model, 26 291 patients with thin melanomas from the US populationbased Surveillance, Epidemiology, and End Results (SEER) cancer registry were used to develop a model that was then validated using 2389 patients seen by the University of Pennsylvania s Pigmented Lesion Group. The model identifies that about 40% of the population representing cases with thin (,1 mm) melanoma will have a 99% predicted probability of 10-year survival. An intermediate group representing 10% of the cases consists of men who have Clark level II/III melanomas with a thickness greater than 0.78 mm or Clark level IV/V melanomas; this group has approximately a 10% probability of death from melanoma. Ulceration is rare in these cases and, when present, is associated with a mortality rate of up to 30%. Since thin melanomas constitute 65% to 80% of melanomas at presentation in today s clinical environment, 19,25 this group of patients, despite having a good survival rate, nevertheless accounts for approximately 15% of the mortality from all melanomas. 25 Interestingly, the thickness cutoff of 0.78 mm, identified in the Gimotty study, 25 is remarkably close to the 0.76 mm cutoff identified by Breslow in his pioneering study. 5,7 Other studies 26,27 have also shown that most metastases from patients with AJCC stage I melanomas occur in the upper quartile of thickness in this subset. In conclusion, prognostication and related clinical decision making for most patients with melanoma can be improved now by using this validated, evidence-based, and peer-reviewed SEER-based classification, 25 which uses readily available prognostic attributes based on histopathologic attributes in common use and understood by all pathologists, as presented in the Table. On the basis of the considerations reviewed above, patients with thin melanomas by Breslow criteria, although considered to be at low risk of mortality overall, should be evaluated by more sophisticated criteria to determine their individual risk. Even though studies of many potential prognostic markers have been conducted, at present, the most highly predictive, appropriately validated studies remain those determined by histopathology with traditional light microscopic methods. The most important of these criteria, mitogenicity, has been incorporated into the new AJCC staging model. Ulceration is also a stage modifier but is rare in thin melanomas. Ki-67 expression, which appears to be independent of mitotic rate as a predictor of survival, 14 is perhaps the most likely candidate marker to be added in the relatively near future. 28,29 Other possibilities that were considered in a recent review include MCAM/MUC18, matrix metalloproteinase 2, proliferating cell nuclear antigen, and p16/ INK4A. 30 The likelihood of introducing these and other markers into general use will be enhanced if their evaluation can be accomplished by an automated counting system, which will need to be based on multiplex labeling immunohistochemistry and quantitative morphometrics. Efforts at developing, validating, and disseminating such systems are in progress. 30,31 References 1. Petersen NC, Bodenham DC, Lloyd OC. Malignant melanomas of the skin. Brit J Plastic Surg. 1962;15:49 94. 2. Mehnert JH, Heard JL. Staging of malignant melanoma by depth of invasion. Am J Surg. 1965;110:168 176. 3. Clark WH Jr. A classification of malignant melanoma in man correlated with histogenesis and biologic behavior. In: Montagna W, Hu F, eds. Advances in the Biology of the Skin. Vol VIII. New York, NY: Pergamon Press; 1967:621 647. 4. Clark WH Jr, From L, Bernardino EA, Mihm MC Jr. The histogenesis and biologic behavior of primary human malignant melanomas of the skin. Cancer Res. 1969;29(3):705 527. 5. Breslow A. Thickness, cross-sectional areas and depth of invasion in the prognosis of cutaneous melanoma. Ann Surg.1970;172(5):902 908. 6. Weinstock MA. Progress and prospects on melanoma: the way forward for early detection and reduced mortality. Clin Cancer Res. 2006;12(7, pt 2):2297s 2300s. 7. Breslow A. Tumor thickness, level of invasion and node dissection in stage I cutaneous melanoma. Ann Surg. 1975;182(5):572 575. 8. Breslow A, Macht SD. Optimal size of resection margin for thin cutaneous melanoma. Surg Gynecol Obstet. 1977;145(5):691 692. 9. Balch CM, Murad TM, Soong SJ, et al. Tumor thickness as a guide to surgical management of clinical stage I melanoma patients. Cancer. 1979;43(3): 883 888. 10. Elder DE, Guerry DIV, Heiberger RM, et al. Optimal resection margin for cutaneous malignant melanoma. Plast Reconstr Surg. 1983;71(1):66 72. 11. Elder DE, Guerry DIV, VanHorn M, et al. The role of lymph node dissection for clinical stage I malignant melanoma of intermediate thickness (1.51 3.99 mm). Cancer. 1985;56(2):413 418. 12. Elder DE, Guerry D, Epstein MN, et al. Invasive malignant melanomas lacking competence for metastasis. Am J Dermatopathol. 1984;(6)(suppl):55 61. 13. Gimotty PA, Guerry D, Ming ME, et al. Thin primary cutaneous malignant melanoma: a prognostic tree for 10-year metastasis is more accurate than American Joint Committee on Cancer staging. J Clin Oncol. 2004;22(18):3668 3676. 14. Gimotty PA, Van BP, Elder DE, et al. Biologic and prognostic significance of dermal Ki67 expression, mitoses, and tumorigenicity in thin invasive cutaneous melanoma. J Clin Oncol. 2005;23(31):8048 8056. 15. Clark WH Jr, Elder DE, Guerry DIV, et al. Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst. 1989;81(24): 1893 1904. 16. Elder DE, Guerry DIV, Epstein MN, et al. Invasive malignant melanomas lacking competence for metastasis. Am J Dermatopathol. 1984;(6)(suppl):55 62. 17. Elder DE, Murphy GF. Malignant tumors (melanomas and related lesions). In: Elder DE, Murphy GF, eds. Melanocytic Tumors of the Skin. Washington, DC: Armed Forces Institute of Pathology; 1991:103 206. Atlas of Tumor Pathology. 3rd series, fascicle 2. Arch Pathol Lab Med Vol 135, March 2011 Thin Melanoma Elder 345

18. 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(16):3635 3648. 19. Gimotty PA, Botbyl J, Soong SJ, Guerry D. A population-based validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol. 2005;23(31):8065 8075. 20. Cochran AJ. Method of assessing prognosis in patients with malignant melanoma. Lancet. 1968;2(7577):1062 1064. 21. Schmoeckel C, Braun-Falco O. Prognostic index in maligant melanoma. Arch Dermatol. 1978;114(6):871 873. 22. Azzola MF, Shaw HM, Thompson JF, et al. Tumor mitotic rate is a more powerful prognostic indicator than ulceration in patients with primary cutaneous melanoma: an analysis of 3661 patients from a single center. Cancer. 2003;97(6): 1488 1498. 23. Scolyer RA, Shaw HM, Thompson JF, et al. Interobserver reproducibility of histopathologic prognostic variables in primary cutaneous melanomas. Am J Surg Pathol. 2003;27(12):1571 1576. 24. Balch CM, Gershenwald JE, Soong SJ, et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol. 2009;27(36):6199 6206. 25. Gimotty PA, Elder DE, Fraker DL, et al. Identification of high-risk patients among those diagnosed with thin cutaneous melanomas. J Clin Oncol. 2007; 25(9):1129 1134. 26. Puleo CA, Messina JL, Riker AI, et al. Sentinel node biopsy for thin melanomas: which patients should be considered? Cancer Control. 2005;12(4): 230 235. 27. Sondak VK, Taylor JM, Sabel MS, et al. Mitotic rate and younger age are predictors of sentinel lymph node positivity: lessons learned from the generation of a probabilistic model. Ann Surg Oncol. 2004;11(3):247 258. 28. Ohsie SJ, Sarantopoulos GP, Cochran AJ, Binder SW. Immunohistochemical characteristics of melanoma. J Cutan Pathol. 2008;35(5):433 444. 29. Pearl RA, Pacifico MD, Richman PI, et al. Ki-67 expression in melanoma: a potential method of risk assessment for the patient with a positive sentinel node. J Exp Clin Cancer Res. 2007;26(1):109 115. 30. Kreizenbeck GM, Berger AJ, Subtil A, et al. Prognostic significance of cadherin-based adhesion molecules in cutaneous malignant melanoma. Cancer Epidemiol Biomarkers Prev. 2008;17(4):949 958. 31. Feldman MD. Beyond morphology: whole slide imaging, computer-aided detection, and other techniques. Arch Pathol Lab Med. 2008;132(5):758 763. 346 Arch Pathol Lab Med Vol 135, March 2011 Thin Melanoma Elder