Biologic and Prognostic Significance of Dermal Ki67 Expression, Mitoses, and Tumorigenicity in Thin Invasive Cutaneous Melanoma
http://www.100md.com
《临床肿瘤学》
the Melanoma Program of the Abramson Cancer Center and the Center for Clinical Epidemiology and Biostatistics, Departments of Biostatistics and Epidemiology, Medicine, Pathology and Laboratory Medicine, and Dermatology, University of Pennsylvania School of Medicine, Philadelphia, PA.
ABSTRACT
PURPOSE: Tumor cell proliferation is a central feature of melanoma progression. In this study, we characterized three biomarkers of proliferation (Ki67 expression, dermal mitotic rate [MR], and tumorigenicity) in thin ( 1.00 mm) primary cutaneous melanomas and examined their association with prognosis.
PATIENTS AND METHODS: We used immunohistochemistry to determine Ki67 expression using the monoclonal antibody MIB-1 in lesions from a prospective cohort that included 396 patients with thin invasive primary melanomas seen between 1972 and 1991. A multivariate Cox proportional hazards model was used to define independent prognostic factors, and recursive partitioning was used to develop a prognostic tree identifying risk groups.
RESULTS: Dermal Ki67 expression was lower than epidermal Ki67 expression in radial growth phase (RGP) melanomas (n = 171), and dermal Ki67 expression and MR were higher in tumorigenic vertical growth phase (VGP) melanomas (n = 193) compared with RGP and nontumorigenic VGP melanomas (n = 42). Dermal Ki67 expression, MR greater than 0, growth phase, thickness, ulceration, tumor-infiltrating lymphocytes, and sex were associated with metastasis at 10 years, however, only dermal Ki67 expression, MR greater than 0, and sex were independent prognostic factors. Two high-risk groups were identified: men and women with dermal MR greater than 0 and dermal Ki67 expression 20% in tumor cells and men with MR greater than 0 and Ki67 expression less than 20%, with 10-year metastasis rates of 39% and 20%, respectively.
CONCLUSION: Proliferation slows as melanoma cells enter the dermis and then increases with the onset of tumorigenic VGP. Ki67 expression and dermal MR provide independent prognostic information that can potentially be used in risk-based management of patients.
INTRODUCTION
Tumor cell proliferation is a key characteristic of stepwise neoplastic progression. The first step in melanoma progression is radial growth phase (RGP), which may be in situ or invasive. In situ radial growth phase is characterized in routine histology by the apparent proliferation of cytologically atypical melanocytes in architecturally disordered patterns (cellular confluence, suprabasal or pagetoid proliferation, with or without mitoses) in the epidermis.1-3 The invasive RGP is characterized by melanocytes that have entered the dermis and that have markers of diminished proliferation (no dermal mitoses and no tumor nests larger than those in the epidermis). In the next step, vertical growth phase (VGP), proliferation is again apparent, as reflected in dermal mitotic figures (mitogenic VGP) and/or tumor cell nests larger than any epidermal nest (tumorigenic VGP). Since its definition in 19892 and modification in 1991,4 this classification scheme has been externally validated,5,6 and we and others have shown that growth phase and dermal mitoses are independent prognostic factors.2,7-10
The majority of patients with melanoma in the United States have thin primary lesions ( 1 mm in thickness).11 The current American Joint Committee on Cancer (AJCC) staging system uses Clark level and ulceration as risk factors for patients with thin lesions, classifying them into two groups: T1a (Clark level II/III tumors without ulceration) and T1b (ulcerated or Clark level IV/V tumors). The 10-year survival rates are 88% and 83% for patients with T1a and T1b melanomas, respectively.12 We recently demonstrated the importance of proliferation-related factors in prognostication, showing that dermal mitoses and growth phase were independent prognostic factors in patients with thin melanomas and that these factors are important in a prognostic tree that was used to identify four risk groups with metastasis-free survival rates that ranged from 70% to 99.5%.13
The Ki67 protein is expressed in all phases of the cell cycle except G0. It therefore has the potential to be a more sensitive biomarker for cellular proliferation than mitoses.14 In melanocytic lesions, Ki67 expression has been proposed as an adjunctive diagnostic biomarker, and its role as a prognostic biomarker has been demonstrated in thicker primary melanomas.15-17 In this study, we focused on three tumor characteristics related to tumor cell proliferation: Ki67 expression, mitotic rate, and growth phase (RGP and tumorigenic and nontumorigenic VGP). In a large cohort of patients with thin lesions and long, prospective follow-up, we addressed five hypotheses related to the biology of tumor progression and to prognosis: (1) tumor cell proliferation as measured by Ki67 expression slows with initial dermal RGP invasion, (2) Ki67 expression increases with the onset of the VGP, (3) dermal tumorigenicity, mitotic rate, and Ki67 expression are each associated with metastasis, (4) Ki67 expression is an independent prognostic factor for metastasis that displaces mitotic rate in multivariable modeling, and (5) a prognostic model using Ki67 expression will be more accurate than previous prognostic models.
PATIENTS AND METHODS
Study Patients and Study Design
The 965 patients with thin, invasive melanomas in this study were identified from a prospectively followed cohort of 1,114 patients seen at the Pigmented Lesion Clinic at the University of Pennsylvania between 1972 and 1991. They were eligible for this study if they had complete data on factors included in the analysis, and, if alive at last contact, had at least 10 years of follow-up.13 Blocks were requested for all patients whose diagnosis was initially made by pathologists at one of three intra- and extra-mural laboratories and located for 411 patients. Because tissue sections of 15 of these 411 blocks were uninterpretable, 396 patients were available for this analysis. These patients were not significantly different from those without blocks (Table 1).
Pathologic Definitions
All slides for routine histologic analysis were initially reviewed by two pathologists without knowledge of patients' outcomes (D.E.E. and W.H. Clark Jr, MD) as described previously2 and a single pathologist (D.E.E.) subsequently rereviewed all specimens to determine tumorigenicity. VGP lesions with at least one dermal cellular nest larger than the largest epidermal nest were classified as tumorigenic. The remaining nontumorigenic VGP lesions had at least one dermal mitosis or a non-nested accretion of melanocytes (the accretive VGP).18,19 Other attributes included dermal mitotic rate (MR), estimated within the region of maximal reactivity and expressed in terms of mitoses per square millimeter (by definition, dermal RGP lesions had no mitoses); VGP tumor infiltrating lymphocytes, classified as either brisk (entirely across the base of the tumorigenic or accretive VGP), nonbrisk, or absent; RGP regression; Breslow thickness; Clark level; microscopic satellites; and ulceration.2
Immunohistochemistry
Expression of the Ki67 antigen was assessed using the immunoglobulin G1 mouse-antihuman monoclonal antibody Ki-67 (clone MIB-1; DAKO, Carpenteria, CA). Antigen retrieval was performed as suggested by the manufacturer but modified as described below, and an automated immunohistochemical staining system (DAKO Cytomation) was used. Formalin-fixed, paraffin-embedded samples were cut at 5 μm, mounted on positively charged microscope slides (Plus slides, Fisher Scientific International Inc, Hampton, NH), and air-dried. Antigen retrieval was accomplished by treatment with heat-induced epitope retrieval buffer (DAKO diluted at 1:10 with sterile water) for two 4-minute cycles of microwave treatment at 70% power. After cooling to room temperature, sections were normalized in a Tris-based buffer for 30 minutes at room temperature. The primary antibody was diluted to a concentration of 1:60, applied to the sections, and incubated for 30 minutes at room temperature. This antibody was detected using a proprietary horseradish peroxidase enzyme-labeled polymer (DAKO Envision+ HRP) conjugated to mouse-secondary antibody and visualized with a stable peroxidase-substrate chromogen, Nova Red (Vector Labs, Burlingame, CA). Sections were counterstained with hematoxylin.
Cells labeled by the antibody displayed a red nuclear staining pattern, except in mitotic cells where the chromosomes and the cytoplasm were labeled. The observers assigned a value of either positive or negative to each melanoma cell of interest. Epidermal and dermal Ki67 expression was assessed by establishing the proportion of positive melanoma cells per 100 cells in each lesional compartment. In lesional compartments within sections (eg, the dermal RGP) having fewer than 100 melanoma cells, the overall proportion of positive cells was recorded for the number of melanoma cells available. Sections were evaluated without knowledge of patients' outcomes by two readers (P.V.B. and T.M. or K.T.M.), and disagreements were resolved by consensus.
Clinical Definitions
Patient characteristics included sex, age, and the location of the melanoma. First metastatic events were classified as follows: regional (in transit dermal or subcutaneous metastases and/or nodal involvement); disseminated (nonregional skin and/or nodes, with or without visceral metastases); or multiple (both regional and disseminated metastasis). Patients with a metastatic event had one or more of the following: a regional or distant metastasis or a melanoma-related death within 10 years of definitive treatment. Local recurrences (within or immediately adjacent to the scar at the primary site) were recorded but not considered metastases. Time to metastasis was the time between definitive treatment and the first metastatic event. Censored patients were those who died from causes unrelated to melanoma or unknown causes and those who were lost to follow-up. Patients without metastasis were followed up for at least 10 years; their mean follow-up was 18.6 years (n = 363 patients). The average time to first metastasis was 6.6 years (n = 33 patients).
Statistical Methods
Pearson's 2 statistic was used to identify differences in characteristics between those included and excluded patients, as well as between patients with tumors in different growth phases. Rank-sum tests were used to assess differences between groups in the percentage of positive melanoma cells. Kaplan-Meier curves were estimated for the time to first metastasis, and the log-rank test was used to identify significant differences between survival curves. Univariate and multivariate Cox proportional hazards regression models were used to obtain unadjusted and adjusted risk ratios presented with their 95% CIs. These analyses were performed using SAS.20 P values are based on two-sided tests except for 2 tests. A classification tree for 10-year metastasis was developed using the recursive partitioning algorithm available in CART (Salford Systems, San Diego, CA).21
The sample size for this study was determined prospectively by considering the relative risks that could be detected with a sample size of 385 patients, assuming an estimate of 10-year risk of metastasis of 0.078, 80% power, and an alpha level of 0.05 at different levels for the prevalence of the risk factors in the population and 10-year risk of metastasis in patients without the risk factor. The minimum value of the detectable relative risk was approximately 3.
RESULTS
Patient and Lesion Characteristics
Characteristics of patients with high- and low-risk melanomas based on Ki67 20% and less than 20%, respectively,22,23 are presented in Table 2. The former were more likely to have primary lesions with adverse prognostic factors. Their lesions were thicker, more often had dermal mitoses and satellites, had higher Clark levels, and were more likely to be tumorigenic.
Characterization of Proliferation Biomarkers
With respect to dermal mitoses, most thin melanomas (73%) had no dermal mitoses, and the mean MR was 0.70 (range, 0 to 13.6). In contrast, Ki67 expression levels were more heterogeneous. Epidermal Ki67 expression appeared normally distributed, with a mean of 23.6% (range, 0% to 80%), whereas dermal Ki67 expression had an asymmetric distribution with a mean of 8.9% (range, 0% to 50%). Only 24% of these melanomas had no dermal Ki67 expression. Approximately half (49%) of the thin melanomas had tumorigenic VGPs (n = 193), and among the 203 remaining nontumorigenic melanomas, 84% were invasive RGP lesions without VGP (n = 171).
A cut point for dermal Ki67 expression associated with 10-year metastasis was determined from the first branch of a classification tree (not shown). Those melanomas with dermal Ki67 expression greater than 20% were associated with a poorer prognosis. Using this cut point, sensitivity and specificity were 48% and 91%, respectively. These values were quite similar to 48% and 89%, the sensitivity and specificity for the cut point used in other publications22,23 where dermal Ki67 expression 20% was used to identify melanomas with poor prognosis. We used dermal Ki67 expression 20% as an unfavorable prognostic factor to be consistent with published work.
A cut point for MR was similarly determined, and lesions with MR greater than 0.3 were associated with a poor prognosis. We evaluated MR greater than 0 as an alternative criterion, and the sensitivity and specificity were exactly the same (76% and 80%, respectively). Consequently, we used MR greater than 0 as an unfavorable prognostic factor.
Proliferation and Progression
For the 171 lesions with invasive RGP but no VGP, epidermal and dermal Ki67 expression differed. Epidermal Ki67 expression was 20% in more than half of these (57%), but only a small percentage (3%) had dermal Ki67 expression 20%. Dermal Ki67 was lower than epidermal Ki67 in all but five of these RGP lesions (Fig 1). The mean of the differences between epidermal and dermal Ki67 expression was 14.8% (standard deviation, 11.5%; paired t statistic, 25.8; P < .001). These findings are consistent with our first hypothesis: proliferation of melanoma cells slows as they enter the dermis.
The next step in melanoma progression is the development of VGP. The distributions of dermal Ki67 expression for the RGPs (n = 171), the nontumorigenic VGPs (n = 32), and the tumorigenic VGPs (n = 193) are presented in Figure 2. There was a significant difference in the proportion of lesions with no dermal Ki67 expression for both nontumorigenic VGP (0%) and tumorigenic VGP (5%) melanomas as compared with RGP melanomas (50.9%; P < .001). In addition, there were significant differences among the means of dermal Ki67 expression for the three groups: 3.5%, 9.3%, and 13.6% in the RGPs, nontumorigenic VGPs, and tumorigenic VGPs, respectively (P < .001). The proportion of melanomas without mitoses among RGP (100%), nontumorigenic VGP (90.6%), and tumorigenic VGP (45%) lesions significantly decreased (P < .001), and the mean dermal MR was higher in tumorigenic VGPs (1.4) compared with nontumorigenic VGPs (0.5; P < .001) and RGPs, where the dermal MR was 0. This pattern is consistent with our second hypothesis: dermal Ki67 expression is higher in VGP than RGP lesions.
Dermal Ki67 Expression and Prognosis
Metastasis rates were higher in melanomas characterized by the presence of biomarkers of proliferation: high dermal Ki67 expression, VGP, the presence of mitoses in the dermis, tumorigenicity, thicker lesions, and Clark level IV, and these patterns were reflected in the associated unadjusted risk ratios from the Cox proportional hazards models (Table 3). The final multivariate regression model included only three prognostic factors: mitogenicity (the presence of any dermal mitoses), dermal Ki67 expression (< 20% or 20%), and sex (Table 4). The adjusted risk ratio of 6.7 for mitogenicity is greater than the adjusted risk ratios for the other two prognostic factors, dermal Ki67 expression and sex (3.4 and 2.5, respectively).
Classification and Prognosis
A prognostic tree (Fig 3) was developed using indicators of poor prognosis (VGP, Clark level IV/V, tumorigenicity, mitogenicity, dermal Ki67 expression 20%, and male sex). Mitogenicity was the first factor selected by the algorithm to identify two groups of patients with differential risk of metastasis. For patients with nonmitogenic melanomas (72%), growth phase was selected as the best prognostic factor (patients with pure, invasive RGP lesions had minimal risk of 10-year metastasis). Among the 28% of patients with mitogenic melanomas, Ki67 expression was selected and patients with low and high dermal Ki67 expression had metastasis rates of 10.3% and 38.7%, respectively. Sex was then selected to identify two groups of patients with a different likelihood of metastasis among those with low dermal Ki67 expression (< 20%), identifying a group of women who had dermal mitoses but low dermal Ki67 expression and a 4.2% rate of 10-year metastasis. Ulceration, tumorigenicity, and Clark level were not selected by the CART algorithm. The survival curves for four risk groups are presented in Figure 4. Although metastatic events have occurred in all groups beyond 10 years after definitive treatment, there remained a significant difference in the likelihood of metastasis between both men and women with mitogenic melanomas with high dermal Ki67 expression (38.7%) and men with mitogenic lesions with low Ki67 expression (20%) compared with those in the other three groups (Fig 4). The sensitivity, specificity, and positive and negative predictive values for several definitions of high risk are presented in Table 5. As hypothesized, the tree-based classification has better diagnostic performance characteristics than other models (including AJCC staging).
DISCUSSION
Based on mitotic counts, it has been thought that tumor progression in melanoma proceeds from the RGP, a phase of rapid tumor cell proliferation in the epidermis that is followed by a phase of invasion and decreased proliferation in the initially inhospitable environment of the dermis, to the VGP, in which higher proliferative rates resume.1-3 In this large study of thin melanomas, we confirmed the observation that the MR varies across tumor progression and found convincing evidence of the same pattern of variation in proliferative rates using a different methodology, the expression of Ki67 as determined by immunohistochemistry. Our study is the first to rigorously quantify these relationships in a large number of thin lesions.
We and others have demonstrated that direct (MR) and indirect (growth phase, level, and thickness) measures of dermal proliferation are prognostic factors in melanomas.2,7-10,12,13,24-30 Since 1988, 65% of invasive cutaneous melanomas reported to the Surveillance, Epidemiology, and End Results Program were thin ( 1 mm).11 In a recent study of the consecutive cohort of patients with thin lesions from which the present study was derived, we demonstrated by multivariate analysis that growth phase, mitogenicity, tumor infiltrating lymphocytes, and sex were independently prognostic.13 Here we tested the hypothesis that measuring the cell cycle–related Ki67 protein would replace mitotic count as a prognostic variable.
Information about the prognostic role of Ki67 expression in patients with thin melanomas is limited. In a study of 167 patients with melanomas less than 1 mm thick, Frahm et al31 found that four of the five patients with a high expression of Ki67 (> 25%) developed metastases, and in a multivariate model, no other factor was independently predictive (age, level, thickness, and mitotic count were tested). Moretti et al32 evaluated Ki67 reactivity in 55 primary melanomas of all thicknesses and found that few patients with thin lesions (< 1.5 mm) had more than 5% positive tumor cells. They found that the minority of their 23 thin lesions were positive and that all of the five positive ones developed metastases, an outcome significantly different from that of the negative lesions. Because thin lesions are prevalent and because most patients with such lesions do not undergo sentinel lymph node biopsies, we have addressed the contribution of Ki67 expression and other measures of proliferation to prognostication in the largest series of such patients yet published.
As expected, dermal Ki67 expression, dermal mitoses, and tumorigenicity were each significantly related to metastasis. However, in the final multivariate model and prognostic tree, only high dermal Ki67 expression ( 20%) and mitogenicity were independent prognostic factors. We had expected dermal mitoses (or MR) to be nonsignificant after adjustment for Ki67 and other prognostic factors. Given the low number of cells in the dermal component of nontumorigenic VGP melanomas and the brief duration of a mitotic figure, it is possible that the longer-lasting Ki67 expression identifies potentially mitogenic cases in which a mitosis does not happen to have been present at the time of excision, whereas the presence of at least a single mitosis retains prognostic value because it is a more unequivocal demonstration of cell division. Capturing proliferation using dermal Ki67 and mitoses on two different sections may also provide additive sources of information.
Although growth phase did not directly enter the final multivariate model as an independent prognostic factor, it did enter the prognostic tree as factor for explaining heterogeneity in metastasis rates among patients with nonmitogenic melanomas. There were 116 VGP melanomas in the nonmitogenic subset, with a 10-year metastasis rates of 2.6, compared with nonmitogenic invasive RGP lesions where this rate was 1.2%.
With prognostic models that better reflect the biology of tumor progression and that more precisely estimate metastatic risk than the current AJCC staging system, patient management will improve. Low-risk patients can be identified and spared intervention beyond surgery at the primary site and be offered less frequent follow-up, largely to detect additional primary lesions. High-risk patients can be identified who are candidates for further staging and therapeutic interventions, including trials of sentinel node biopsy and adjuvant therapy. The prognostic tree presented here identifies patient groups whose 10-year metastatic rate varied from 1.2% to 38.7%. This classification suggests that approximately one quarter of patients with thin melanomas (those with mitogenic melanomas) should have their lesions evaluated for Ki67 expression. Women with mitogenic melanomas but low dermal Ki67 expression (21% of all women and 12% of all patients) might not be candidates for sentinel node biopsy. It also suggests that Ki67 expression identifies a high-risk group for adjuvant therapy trials. As noted in Table 5, mitogenicity and Ki67 outperformed other measures of risk, including level and ulceration.
These considerations indicate that it will be important to validate these results in another data set. However, the present study has a number of strengths that suggest that it is robust. The study cohort comprised patients with available paraffin blocks who were members of a prospectively defined inception cohort that was rigorously followed up for more than 10 years. Their demographics and routine pathology attributes were not appreciably different from those of patients not included in this study because of unavailable blocks. In addition, the immunohistologic prognostic biomarker was carefully evaluated for its potential use in clinical practice. As with any candidate immunohistochemical biomarker, for Ki67 expression to become a clinically useful biomarker in melanoma, it will be important to standardize its assessment in both thin and thick primary lesions33,34
In this study we have described Ki67 expression in thin melanomas within the context of melanoma progression, demonstrated its association with metastasis and its role as an independent prognostic factor, and illustrated its practical application as a complement to a more established biomarker based on dermal mitoses. We have also demonstrated that MR conveniently evaluated as a binary attribute (present or absent) is prognostic, greatly simplifying its recording by pathologists. These results, in light of recent studies identifying a role for MR in prognosis and prediction for sentinel lymph node biopsies, suggest that Ki67 expression should be considered as an adjunct biomarker to mitogenicity. Our findings need to be replicated in future prospective studies of clinical outcomes, including the prediction of nodal status.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all the patients who have been seen at the Pigmented Lesion Clinic (PLC) and gave their consent for use of their data in research studies, as well as the investigators (Dr. W.H. Clark Jr [deceased],E.E. Bondi, L.P. Bucky, L.S. Callans, B. Chang, K.T. Flaherty, D.L. Fraker, A.C. Halpern, R. Hamilton, D. Hershock, D.D. Larossa, S.R. Lessin, D. Low, P. Van Belle, and J. Wolfe) and staff (R. Holmes, S. Hotz, N. Lowden, I. Matozzo, M. Price, M. Synnestvedt, and J. Thompson) of the PLC for their contributions over the last three decades to the Melanoma Core Database on which this report is based.
NOTES
Supported in part by Grant Nos. CA-75434, CA-25874, and CA-093372 from the National Institutes of Health and the National Cancer Institute, Bethesda MD.
Presented in part at the 93rd Annual Meeting of the American Association for Cancer Research, April 5-11, 2002, San Francisco, CA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Clark WH, From L, Bernardino EA, et al: The histogenesis and biologic behavior of primary human malignant melanomas of the skin. Cancer Res 29:705-727, 1969
Clark WH, Elder DE, Guerry D, et al: Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst 81:1893-1904, 1989
Guerry D, Synnestvedt M, Elder DE, et al: Lessons learned from tumor progression: The invasive radial growth phase of melanoma is common, incapable of metastasis, and indolent. J Invest Dermatol 100:342S–345S, 1993
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, pp 103-154
Cook MG, Clarke TJ, Humphreys S, et al: A nationwide survey of observer variation in the diagnosis of thin cutaneous malignant melanoma including the MIN terminology. J Clin Pathol 50:202-205, 1997
McDermott NC, Hayes DP, al-Sader MH, et al: Identification of vertical growth phase in malignant melanoma: A study of interobserver agreement. Am J Clin Pathol 110:753-757, 1998
Oliveira Filho RS, Ferreira LM, Biasi LJ, et al: Vertical growth phase and positive sentinel node in thin melanoma. Braz J Med Biol Res 36:347-350, 2003
Cook MG, Spatz A, Brocker E, et al: Identification of histological features associated with metastatic potential of thin (< 1.0 mm) cutaneous melanoma with metastases: A study on behalf of the EORTC Melanoma Group. J Pathol 197:188-193, 2002
Lefevre M, Vergier B, Balme B, et al: Relevance of vertical growth pattern in thin level II cutaneous superficial spreading melanomas. Am J Surg Pathol 27:717-724, 2003
Massi D, Franchi A, Borgognoni L, et al: Thin cutaneous malignant melanomas (1.5 mm): Identification of risk factors indicative of progression. Cancer 85:1067-1076, 1999
Gimotty PA, Botbyl J, Soong SJ, et al: A population-based validation of the AJCC staging system for melanoma. J Clin Oncol (in press)
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 19:3635-3648, 2001
Gimotty PA, Guerry D, Ming M, 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 22:3668-3676, 2004
Brown DC, Gatter KC: Ki67 protein: The immaculate deception? Histopathology 40:2-11, 2002
de Wit NJW, van Muijen GNP, Ruiter DJ: Immunohistochemistry in melanocytic proliferative lesions. Histopathology 44:517-541, 2004
Lomuto M, Calabrese P, Giuliani A: Prognostic signs in melanoma: State of the art. J Eur Acad Dermatol Venereol 18:291-300, 2004
Li N, Mangini J, Bhawan J: New prognostic factors of cutaneous melanoma: A review of the literature. J Cutan Pathol 29:324-340, 2002
Elder DE, Van Belle PA, Elenitsas R, et al: Neoplastic progression and prognosis in melanoma. Semin Cutan Med Surg 15:336-348, 1996
Reed RJ: Minimal deviation melanoma, in Mihm MC, Murphy GF, Kaufman N (eds): Pathobiology and Recognition of Malignant Melanoma. Baltimore, MD, Williams & Wilkins, 1988, pp 110-152
SAS Institute Inc: SAS/STAT User's Guide, Version 9.1. Cary, NC, SAS Institute Inc, 2004
Steinberg D, Colla P: CART: Tree-structured non-parametric data analysis. San Diego, CA, Salford Systems, 1995
Hazan C, Metzer K, Panageas KS, et al: Evaluation of the proliferation marker MIB-1 in the prognosis of cutaneous malignant melanoma. Cancer 95:634-640, 2002
Alonso SR, Ortiz P, Pollan M, et al: Progression in cutaneous malignant melanoma is associated with distinct expression profiles: A tissue microarray-based study. Am J Pathol 164:193-203, 2004
Balch CM, Soong SJ, Gershenwald JE, et al: Prognostic factors analysis of 17,600 melanoma patients: Validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 19:3622-3634, 2001
McKinnon JG, Yu XQ, McCarthy WH, et al: Prognosis for patients with thin cutaneous melanoma. Cancer 98:1223-1231, 2003
Mansson-Brahme E, Carstensen J, Erhardt K, et al: Prognostic factors in thin cutaneous malignant melanoma. Cancer 73:2324-2332, 1994
Finley JW, Gibbs JF, Rodriguez LM, et al: Pathologic and clinical features influencing outcome of thin cutaneous melanoma: Correlation with newly proposed staging system. Am Surg 66:527-531, 2000
Ostmeier H, Fuchs B, Otto F, et al: Can immunohistochemical markers and mitotic rate improve prediction in patients with primary melanoma? Cancer 85:2391-2399, 1999
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. Cancer 97:1488-1498, 2003
Francken AB, Shaw HM, Thompson JF, et al: The prognostic importance of tumor mitotic rate confirmed in 1317 patients with primary cutaneous melanoma and long follow-up. Ann Surg Oncol 11:426-433, 2004
Frahm S, Schubert C, Parwaresch R, et al: High proliferative activity may predict early metastasis of thin melanomas. Hum Pathol 32:1376-1381, 2001
Moretti S, Spallanzani A, Chiarugi A, et al: Correlation of Ki67 expression in cutaneous primary melanoma with prognosis in a prospective study: Different correlation according to thickness. J Am Acad Dermatol 44:188-192, 2001
Pepe MS, Etzioni R, Feng Z, et al: Phases of biomarker development for early detection of cancer. J Natl Cancer Inst 93:1054-1061, 2001
Barker PE: Cancer biomarker validation: Standards and process. Ann N Y Acad Sci 983:142-150, 2003
Submitted March 20, 2005; accepted June 28, 2005.(Phyllis A. Gimotty, Patri)
ABSTRACT
PURPOSE: Tumor cell proliferation is a central feature of melanoma progression. In this study, we characterized three biomarkers of proliferation (Ki67 expression, dermal mitotic rate [MR], and tumorigenicity) in thin ( 1.00 mm) primary cutaneous melanomas and examined their association with prognosis.
PATIENTS AND METHODS: We used immunohistochemistry to determine Ki67 expression using the monoclonal antibody MIB-1 in lesions from a prospective cohort that included 396 patients with thin invasive primary melanomas seen between 1972 and 1991. A multivariate Cox proportional hazards model was used to define independent prognostic factors, and recursive partitioning was used to develop a prognostic tree identifying risk groups.
RESULTS: Dermal Ki67 expression was lower than epidermal Ki67 expression in radial growth phase (RGP) melanomas (n = 171), and dermal Ki67 expression and MR were higher in tumorigenic vertical growth phase (VGP) melanomas (n = 193) compared with RGP and nontumorigenic VGP melanomas (n = 42). Dermal Ki67 expression, MR greater than 0, growth phase, thickness, ulceration, tumor-infiltrating lymphocytes, and sex were associated with metastasis at 10 years, however, only dermal Ki67 expression, MR greater than 0, and sex were independent prognostic factors. Two high-risk groups were identified: men and women with dermal MR greater than 0 and dermal Ki67 expression 20% in tumor cells and men with MR greater than 0 and Ki67 expression less than 20%, with 10-year metastasis rates of 39% and 20%, respectively.
CONCLUSION: Proliferation slows as melanoma cells enter the dermis and then increases with the onset of tumorigenic VGP. Ki67 expression and dermal MR provide independent prognostic information that can potentially be used in risk-based management of patients.
INTRODUCTION
Tumor cell proliferation is a key characteristic of stepwise neoplastic progression. The first step in melanoma progression is radial growth phase (RGP), which may be in situ or invasive. In situ radial growth phase is characterized in routine histology by the apparent proliferation of cytologically atypical melanocytes in architecturally disordered patterns (cellular confluence, suprabasal or pagetoid proliferation, with or without mitoses) in the epidermis.1-3 The invasive RGP is characterized by melanocytes that have entered the dermis and that have markers of diminished proliferation (no dermal mitoses and no tumor nests larger than those in the epidermis). In the next step, vertical growth phase (VGP), proliferation is again apparent, as reflected in dermal mitotic figures (mitogenic VGP) and/or tumor cell nests larger than any epidermal nest (tumorigenic VGP). Since its definition in 19892 and modification in 1991,4 this classification scheme has been externally validated,5,6 and we and others have shown that growth phase and dermal mitoses are independent prognostic factors.2,7-10
The majority of patients with melanoma in the United States have thin primary lesions ( 1 mm in thickness).11 The current American Joint Committee on Cancer (AJCC) staging system uses Clark level and ulceration as risk factors for patients with thin lesions, classifying them into two groups: T1a (Clark level II/III tumors without ulceration) and T1b (ulcerated or Clark level IV/V tumors). The 10-year survival rates are 88% and 83% for patients with T1a and T1b melanomas, respectively.12 We recently demonstrated the importance of proliferation-related factors in prognostication, showing that dermal mitoses and growth phase were independent prognostic factors in patients with thin melanomas and that these factors are important in a prognostic tree that was used to identify four risk groups with metastasis-free survival rates that ranged from 70% to 99.5%.13
The Ki67 protein is expressed in all phases of the cell cycle except G0. It therefore has the potential to be a more sensitive biomarker for cellular proliferation than mitoses.14 In melanocytic lesions, Ki67 expression has been proposed as an adjunctive diagnostic biomarker, and its role as a prognostic biomarker has been demonstrated in thicker primary melanomas.15-17 In this study, we focused on three tumor characteristics related to tumor cell proliferation: Ki67 expression, mitotic rate, and growth phase (RGP and tumorigenic and nontumorigenic VGP). In a large cohort of patients with thin lesions and long, prospective follow-up, we addressed five hypotheses related to the biology of tumor progression and to prognosis: (1) tumor cell proliferation as measured by Ki67 expression slows with initial dermal RGP invasion, (2) Ki67 expression increases with the onset of the VGP, (3) dermal tumorigenicity, mitotic rate, and Ki67 expression are each associated with metastasis, (4) Ki67 expression is an independent prognostic factor for metastasis that displaces mitotic rate in multivariable modeling, and (5) a prognostic model using Ki67 expression will be more accurate than previous prognostic models.
PATIENTS AND METHODS
Study Patients and Study Design
The 965 patients with thin, invasive melanomas in this study were identified from a prospectively followed cohort of 1,114 patients seen at the Pigmented Lesion Clinic at the University of Pennsylvania between 1972 and 1991. They were eligible for this study if they had complete data on factors included in the analysis, and, if alive at last contact, had at least 10 years of follow-up.13 Blocks were requested for all patients whose diagnosis was initially made by pathologists at one of three intra- and extra-mural laboratories and located for 411 patients. Because tissue sections of 15 of these 411 blocks were uninterpretable, 396 patients were available for this analysis. These patients were not significantly different from those without blocks (Table 1).
Pathologic Definitions
All slides for routine histologic analysis were initially reviewed by two pathologists without knowledge of patients' outcomes (D.E.E. and W.H. Clark Jr, MD) as described previously2 and a single pathologist (D.E.E.) subsequently rereviewed all specimens to determine tumorigenicity. VGP lesions with at least one dermal cellular nest larger than the largest epidermal nest were classified as tumorigenic. The remaining nontumorigenic VGP lesions had at least one dermal mitosis or a non-nested accretion of melanocytes (the accretive VGP).18,19 Other attributes included dermal mitotic rate (MR), estimated within the region of maximal reactivity and expressed in terms of mitoses per square millimeter (by definition, dermal RGP lesions had no mitoses); VGP tumor infiltrating lymphocytes, classified as either brisk (entirely across the base of the tumorigenic or accretive VGP), nonbrisk, or absent; RGP regression; Breslow thickness; Clark level; microscopic satellites; and ulceration.2
Immunohistochemistry
Expression of the Ki67 antigen was assessed using the immunoglobulin G1 mouse-antihuman monoclonal antibody Ki-67 (clone MIB-1; DAKO, Carpenteria, CA). Antigen retrieval was performed as suggested by the manufacturer but modified as described below, and an automated immunohistochemical staining system (DAKO Cytomation) was used. Formalin-fixed, paraffin-embedded samples were cut at 5 μm, mounted on positively charged microscope slides (Plus slides, Fisher Scientific International Inc, Hampton, NH), and air-dried. Antigen retrieval was accomplished by treatment with heat-induced epitope retrieval buffer (DAKO diluted at 1:10 with sterile water) for two 4-minute cycles of microwave treatment at 70% power. After cooling to room temperature, sections were normalized in a Tris-based buffer for 30 minutes at room temperature. The primary antibody was diluted to a concentration of 1:60, applied to the sections, and incubated for 30 minutes at room temperature. This antibody was detected using a proprietary horseradish peroxidase enzyme-labeled polymer (DAKO Envision+ HRP) conjugated to mouse-secondary antibody and visualized with a stable peroxidase-substrate chromogen, Nova Red (Vector Labs, Burlingame, CA). Sections were counterstained with hematoxylin.
Cells labeled by the antibody displayed a red nuclear staining pattern, except in mitotic cells where the chromosomes and the cytoplasm were labeled. The observers assigned a value of either positive or negative to each melanoma cell of interest. Epidermal and dermal Ki67 expression was assessed by establishing the proportion of positive melanoma cells per 100 cells in each lesional compartment. In lesional compartments within sections (eg, the dermal RGP) having fewer than 100 melanoma cells, the overall proportion of positive cells was recorded for the number of melanoma cells available. Sections were evaluated without knowledge of patients' outcomes by two readers (P.V.B. and T.M. or K.T.M.), and disagreements were resolved by consensus.
Clinical Definitions
Patient characteristics included sex, age, and the location of the melanoma. First metastatic events were classified as follows: regional (in transit dermal or subcutaneous metastases and/or nodal involvement); disseminated (nonregional skin and/or nodes, with or without visceral metastases); or multiple (both regional and disseminated metastasis). Patients with a metastatic event had one or more of the following: a regional or distant metastasis or a melanoma-related death within 10 years of definitive treatment. Local recurrences (within or immediately adjacent to the scar at the primary site) were recorded but not considered metastases. Time to metastasis was the time between definitive treatment and the first metastatic event. Censored patients were those who died from causes unrelated to melanoma or unknown causes and those who were lost to follow-up. Patients without metastasis were followed up for at least 10 years; their mean follow-up was 18.6 years (n = 363 patients). The average time to first metastasis was 6.6 years (n = 33 patients).
Statistical Methods
Pearson's 2 statistic was used to identify differences in characteristics between those included and excluded patients, as well as between patients with tumors in different growth phases. Rank-sum tests were used to assess differences between groups in the percentage of positive melanoma cells. Kaplan-Meier curves were estimated for the time to first metastasis, and the log-rank test was used to identify significant differences between survival curves. Univariate and multivariate Cox proportional hazards regression models were used to obtain unadjusted and adjusted risk ratios presented with their 95% CIs. These analyses were performed using SAS.20 P values are based on two-sided tests except for 2 tests. A classification tree for 10-year metastasis was developed using the recursive partitioning algorithm available in CART (Salford Systems, San Diego, CA).21
The sample size for this study was determined prospectively by considering the relative risks that could be detected with a sample size of 385 patients, assuming an estimate of 10-year risk of metastasis of 0.078, 80% power, and an alpha level of 0.05 at different levels for the prevalence of the risk factors in the population and 10-year risk of metastasis in patients without the risk factor. The minimum value of the detectable relative risk was approximately 3.
RESULTS
Patient and Lesion Characteristics
Characteristics of patients with high- and low-risk melanomas based on Ki67 20% and less than 20%, respectively,22,23 are presented in Table 2. The former were more likely to have primary lesions with adverse prognostic factors. Their lesions were thicker, more often had dermal mitoses and satellites, had higher Clark levels, and were more likely to be tumorigenic.
Characterization of Proliferation Biomarkers
With respect to dermal mitoses, most thin melanomas (73%) had no dermal mitoses, and the mean MR was 0.70 (range, 0 to 13.6). In contrast, Ki67 expression levels were more heterogeneous. Epidermal Ki67 expression appeared normally distributed, with a mean of 23.6% (range, 0% to 80%), whereas dermal Ki67 expression had an asymmetric distribution with a mean of 8.9% (range, 0% to 50%). Only 24% of these melanomas had no dermal Ki67 expression. Approximately half (49%) of the thin melanomas had tumorigenic VGPs (n = 193), and among the 203 remaining nontumorigenic melanomas, 84% were invasive RGP lesions without VGP (n = 171).
A cut point for dermal Ki67 expression associated with 10-year metastasis was determined from the first branch of a classification tree (not shown). Those melanomas with dermal Ki67 expression greater than 20% were associated with a poorer prognosis. Using this cut point, sensitivity and specificity were 48% and 91%, respectively. These values were quite similar to 48% and 89%, the sensitivity and specificity for the cut point used in other publications22,23 where dermal Ki67 expression 20% was used to identify melanomas with poor prognosis. We used dermal Ki67 expression 20% as an unfavorable prognostic factor to be consistent with published work.
A cut point for MR was similarly determined, and lesions with MR greater than 0.3 were associated with a poor prognosis. We evaluated MR greater than 0 as an alternative criterion, and the sensitivity and specificity were exactly the same (76% and 80%, respectively). Consequently, we used MR greater than 0 as an unfavorable prognostic factor.
Proliferation and Progression
For the 171 lesions with invasive RGP but no VGP, epidermal and dermal Ki67 expression differed. Epidermal Ki67 expression was 20% in more than half of these (57%), but only a small percentage (3%) had dermal Ki67 expression 20%. Dermal Ki67 was lower than epidermal Ki67 in all but five of these RGP lesions (Fig 1). The mean of the differences between epidermal and dermal Ki67 expression was 14.8% (standard deviation, 11.5%; paired t statistic, 25.8; P < .001). These findings are consistent with our first hypothesis: proliferation of melanoma cells slows as they enter the dermis.
The next step in melanoma progression is the development of VGP. The distributions of dermal Ki67 expression for the RGPs (n = 171), the nontumorigenic VGPs (n = 32), and the tumorigenic VGPs (n = 193) are presented in Figure 2. There was a significant difference in the proportion of lesions with no dermal Ki67 expression for both nontumorigenic VGP (0%) and tumorigenic VGP (5%) melanomas as compared with RGP melanomas (50.9%; P < .001). In addition, there were significant differences among the means of dermal Ki67 expression for the three groups: 3.5%, 9.3%, and 13.6% in the RGPs, nontumorigenic VGPs, and tumorigenic VGPs, respectively (P < .001). The proportion of melanomas without mitoses among RGP (100%), nontumorigenic VGP (90.6%), and tumorigenic VGP (45%) lesions significantly decreased (P < .001), and the mean dermal MR was higher in tumorigenic VGPs (1.4) compared with nontumorigenic VGPs (0.5; P < .001) and RGPs, where the dermal MR was 0. This pattern is consistent with our second hypothesis: dermal Ki67 expression is higher in VGP than RGP lesions.
Dermal Ki67 Expression and Prognosis
Metastasis rates were higher in melanomas characterized by the presence of biomarkers of proliferation: high dermal Ki67 expression, VGP, the presence of mitoses in the dermis, tumorigenicity, thicker lesions, and Clark level IV, and these patterns were reflected in the associated unadjusted risk ratios from the Cox proportional hazards models (Table 3). The final multivariate regression model included only three prognostic factors: mitogenicity (the presence of any dermal mitoses), dermal Ki67 expression (< 20% or 20%), and sex (Table 4). The adjusted risk ratio of 6.7 for mitogenicity is greater than the adjusted risk ratios for the other two prognostic factors, dermal Ki67 expression and sex (3.4 and 2.5, respectively).
Classification and Prognosis
A prognostic tree (Fig 3) was developed using indicators of poor prognosis (VGP, Clark level IV/V, tumorigenicity, mitogenicity, dermal Ki67 expression 20%, and male sex). Mitogenicity was the first factor selected by the algorithm to identify two groups of patients with differential risk of metastasis. For patients with nonmitogenic melanomas (72%), growth phase was selected as the best prognostic factor (patients with pure, invasive RGP lesions had minimal risk of 10-year metastasis). Among the 28% of patients with mitogenic melanomas, Ki67 expression was selected and patients with low and high dermal Ki67 expression had metastasis rates of 10.3% and 38.7%, respectively. Sex was then selected to identify two groups of patients with a different likelihood of metastasis among those with low dermal Ki67 expression (< 20%), identifying a group of women who had dermal mitoses but low dermal Ki67 expression and a 4.2% rate of 10-year metastasis. Ulceration, tumorigenicity, and Clark level were not selected by the CART algorithm. The survival curves for four risk groups are presented in Figure 4. Although metastatic events have occurred in all groups beyond 10 years after definitive treatment, there remained a significant difference in the likelihood of metastasis between both men and women with mitogenic melanomas with high dermal Ki67 expression (38.7%) and men with mitogenic lesions with low Ki67 expression (20%) compared with those in the other three groups (Fig 4). The sensitivity, specificity, and positive and negative predictive values for several definitions of high risk are presented in Table 5. As hypothesized, the tree-based classification has better diagnostic performance characteristics than other models (including AJCC staging).
DISCUSSION
Based on mitotic counts, it has been thought that tumor progression in melanoma proceeds from the RGP, a phase of rapid tumor cell proliferation in the epidermis that is followed by a phase of invasion and decreased proliferation in the initially inhospitable environment of the dermis, to the VGP, in which higher proliferative rates resume.1-3 In this large study of thin melanomas, we confirmed the observation that the MR varies across tumor progression and found convincing evidence of the same pattern of variation in proliferative rates using a different methodology, the expression of Ki67 as determined by immunohistochemistry. Our study is the first to rigorously quantify these relationships in a large number of thin lesions.
We and others have demonstrated that direct (MR) and indirect (growth phase, level, and thickness) measures of dermal proliferation are prognostic factors in melanomas.2,7-10,12,13,24-30 Since 1988, 65% of invasive cutaneous melanomas reported to the Surveillance, Epidemiology, and End Results Program were thin ( 1 mm).11 In a recent study of the consecutive cohort of patients with thin lesions from which the present study was derived, we demonstrated by multivariate analysis that growth phase, mitogenicity, tumor infiltrating lymphocytes, and sex were independently prognostic.13 Here we tested the hypothesis that measuring the cell cycle–related Ki67 protein would replace mitotic count as a prognostic variable.
Information about the prognostic role of Ki67 expression in patients with thin melanomas is limited. In a study of 167 patients with melanomas less than 1 mm thick, Frahm et al31 found that four of the five patients with a high expression of Ki67 (> 25%) developed metastases, and in a multivariate model, no other factor was independently predictive (age, level, thickness, and mitotic count were tested). Moretti et al32 evaluated Ki67 reactivity in 55 primary melanomas of all thicknesses and found that few patients with thin lesions (< 1.5 mm) had more than 5% positive tumor cells. They found that the minority of their 23 thin lesions were positive and that all of the five positive ones developed metastases, an outcome significantly different from that of the negative lesions. Because thin lesions are prevalent and because most patients with such lesions do not undergo sentinel lymph node biopsies, we have addressed the contribution of Ki67 expression and other measures of proliferation to prognostication in the largest series of such patients yet published.
As expected, dermal Ki67 expression, dermal mitoses, and tumorigenicity were each significantly related to metastasis. However, in the final multivariate model and prognostic tree, only high dermal Ki67 expression ( 20%) and mitogenicity were independent prognostic factors. We had expected dermal mitoses (or MR) to be nonsignificant after adjustment for Ki67 and other prognostic factors. Given the low number of cells in the dermal component of nontumorigenic VGP melanomas and the brief duration of a mitotic figure, it is possible that the longer-lasting Ki67 expression identifies potentially mitogenic cases in which a mitosis does not happen to have been present at the time of excision, whereas the presence of at least a single mitosis retains prognostic value because it is a more unequivocal demonstration of cell division. Capturing proliferation using dermal Ki67 and mitoses on two different sections may also provide additive sources of information.
Although growth phase did not directly enter the final multivariate model as an independent prognostic factor, it did enter the prognostic tree as factor for explaining heterogeneity in metastasis rates among patients with nonmitogenic melanomas. There were 116 VGP melanomas in the nonmitogenic subset, with a 10-year metastasis rates of 2.6, compared with nonmitogenic invasive RGP lesions where this rate was 1.2%.
With prognostic models that better reflect the biology of tumor progression and that more precisely estimate metastatic risk than the current AJCC staging system, patient management will improve. Low-risk patients can be identified and spared intervention beyond surgery at the primary site and be offered less frequent follow-up, largely to detect additional primary lesions. High-risk patients can be identified who are candidates for further staging and therapeutic interventions, including trials of sentinel node biopsy and adjuvant therapy. The prognostic tree presented here identifies patient groups whose 10-year metastatic rate varied from 1.2% to 38.7%. This classification suggests that approximately one quarter of patients with thin melanomas (those with mitogenic melanomas) should have their lesions evaluated for Ki67 expression. Women with mitogenic melanomas but low dermal Ki67 expression (21% of all women and 12% of all patients) might not be candidates for sentinel node biopsy. It also suggests that Ki67 expression identifies a high-risk group for adjuvant therapy trials. As noted in Table 5, mitogenicity and Ki67 outperformed other measures of risk, including level and ulceration.
These considerations indicate that it will be important to validate these results in another data set. However, the present study has a number of strengths that suggest that it is robust. The study cohort comprised patients with available paraffin blocks who were members of a prospectively defined inception cohort that was rigorously followed up for more than 10 years. Their demographics and routine pathology attributes were not appreciably different from those of patients not included in this study because of unavailable blocks. In addition, the immunohistologic prognostic biomarker was carefully evaluated for its potential use in clinical practice. As with any candidate immunohistochemical biomarker, for Ki67 expression to become a clinically useful biomarker in melanoma, it will be important to standardize its assessment in both thin and thick primary lesions33,34
In this study we have described Ki67 expression in thin melanomas within the context of melanoma progression, demonstrated its association with metastasis and its role as an independent prognostic factor, and illustrated its practical application as a complement to a more established biomarker based on dermal mitoses. We have also demonstrated that MR conveniently evaluated as a binary attribute (present or absent) is prognostic, greatly simplifying its recording by pathologists. These results, in light of recent studies identifying a role for MR in prognosis and prediction for sentinel lymph node biopsies, suggest that Ki67 expression should be considered as an adjunct biomarker to mitogenicity. Our findings need to be replicated in future prospective studies of clinical outcomes, including the prediction of nodal status.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all the patients who have been seen at the Pigmented Lesion Clinic (PLC) and gave their consent for use of their data in research studies, as well as the investigators (Dr. W.H. Clark Jr [deceased],E.E. Bondi, L.P. Bucky, L.S. Callans, B. Chang, K.T. Flaherty, D.L. Fraker, A.C. Halpern, R. Hamilton, D. Hershock, D.D. Larossa, S.R. Lessin, D. Low, P. Van Belle, and J. Wolfe) and staff (R. Holmes, S. Hotz, N. Lowden, I. Matozzo, M. Price, M. Synnestvedt, and J. Thompson) of the PLC for their contributions over the last three decades to the Melanoma Core Database on which this report is based.
NOTES
Supported in part by Grant Nos. CA-75434, CA-25874, and CA-093372 from the National Institutes of Health and the National Cancer Institute, Bethesda MD.
Presented in part at the 93rd Annual Meeting of the American Association for Cancer Research, April 5-11, 2002, San Francisco, CA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Clark WH, From L, Bernardino EA, et al: The histogenesis and biologic behavior of primary human malignant melanomas of the skin. Cancer Res 29:705-727, 1969
Clark WH, Elder DE, Guerry D, et al: Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst 81:1893-1904, 1989
Guerry D, Synnestvedt M, Elder DE, et al: Lessons learned from tumor progression: The invasive radial growth phase of melanoma is common, incapable of metastasis, and indolent. J Invest Dermatol 100:342S–345S, 1993
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, pp 103-154
Cook MG, Clarke TJ, Humphreys S, et al: A nationwide survey of observer variation in the diagnosis of thin cutaneous malignant melanoma including the MIN terminology. J Clin Pathol 50:202-205, 1997
McDermott NC, Hayes DP, al-Sader MH, et al: Identification of vertical growth phase in malignant melanoma: A study of interobserver agreement. Am J Clin Pathol 110:753-757, 1998
Oliveira Filho RS, Ferreira LM, Biasi LJ, et al: Vertical growth phase and positive sentinel node in thin melanoma. Braz J Med Biol Res 36:347-350, 2003
Cook MG, Spatz A, Brocker E, et al: Identification of histological features associated with metastatic potential of thin (< 1.0 mm) cutaneous melanoma with metastases: A study on behalf of the EORTC Melanoma Group. J Pathol 197:188-193, 2002
Lefevre M, Vergier B, Balme B, et al: Relevance of vertical growth pattern in thin level II cutaneous superficial spreading melanomas. Am J Surg Pathol 27:717-724, 2003
Massi D, Franchi A, Borgognoni L, et al: Thin cutaneous malignant melanomas (1.5 mm): Identification of risk factors indicative of progression. Cancer 85:1067-1076, 1999
Gimotty PA, Botbyl J, Soong SJ, et al: A population-based validation of the AJCC staging system for melanoma. J Clin Oncol (in press)
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 19:3635-3648, 2001
Gimotty PA, Guerry D, Ming M, 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 22:3668-3676, 2004
Brown DC, Gatter KC: Ki67 protein: The immaculate deception? Histopathology 40:2-11, 2002
de Wit NJW, van Muijen GNP, Ruiter DJ: Immunohistochemistry in melanocytic proliferative lesions. Histopathology 44:517-541, 2004
Lomuto M, Calabrese P, Giuliani A: Prognostic signs in melanoma: State of the art. J Eur Acad Dermatol Venereol 18:291-300, 2004
Li N, Mangini J, Bhawan J: New prognostic factors of cutaneous melanoma: A review of the literature. J Cutan Pathol 29:324-340, 2002
Elder DE, Van Belle PA, Elenitsas R, et al: Neoplastic progression and prognosis in melanoma. Semin Cutan Med Surg 15:336-348, 1996
Reed RJ: Minimal deviation melanoma, in Mihm MC, Murphy GF, Kaufman N (eds): Pathobiology and Recognition of Malignant Melanoma. Baltimore, MD, Williams & Wilkins, 1988, pp 110-152
SAS Institute Inc: SAS/STAT User's Guide, Version 9.1. Cary, NC, SAS Institute Inc, 2004
Steinberg D, Colla P: CART: Tree-structured non-parametric data analysis. San Diego, CA, Salford Systems, 1995
Hazan C, Metzer K, Panageas KS, et al: Evaluation of the proliferation marker MIB-1 in the prognosis of cutaneous malignant melanoma. Cancer 95:634-640, 2002
Alonso SR, Ortiz P, Pollan M, et al: Progression in cutaneous malignant melanoma is associated with distinct expression profiles: A tissue microarray-based study. Am J Pathol 164:193-203, 2004
Balch CM, Soong SJ, Gershenwald JE, et al: Prognostic factors analysis of 17,600 melanoma patients: Validation of the American Joint Committee on Cancer melanoma staging system. J Clin Oncol 19:3622-3634, 2001
McKinnon JG, Yu XQ, McCarthy WH, et al: Prognosis for patients with thin cutaneous melanoma. Cancer 98:1223-1231, 2003
Mansson-Brahme E, Carstensen J, Erhardt K, et al: Prognostic factors in thin cutaneous malignant melanoma. Cancer 73:2324-2332, 1994
Finley JW, Gibbs JF, Rodriguez LM, et al: Pathologic and clinical features influencing outcome of thin cutaneous melanoma: Correlation with newly proposed staging system. Am Surg 66:527-531, 2000
Ostmeier H, Fuchs B, Otto F, et al: Can immunohistochemical markers and mitotic rate improve prediction in patients with primary melanoma? Cancer 85:2391-2399, 1999
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. Cancer 97:1488-1498, 2003
Francken AB, Shaw HM, Thompson JF, et al: The prognostic importance of tumor mitotic rate confirmed in 1317 patients with primary cutaneous melanoma and long follow-up. Ann Surg Oncol 11:426-433, 2004
Frahm S, Schubert C, Parwaresch R, et al: High proliferative activity may predict early metastasis of thin melanomas. Hum Pathol 32:1376-1381, 2001
Moretti S, Spallanzani A, Chiarugi A, et al: Correlation of Ki67 expression in cutaneous primary melanoma with prognosis in a prospective study: Different correlation according to thickness. J Am Acad Dermatol 44:188-192, 2001
Pepe MS, Etzioni R, Feng Z, et al: Phases of biomarker development for early detection of cancer. J Natl Cancer Inst 93:1054-1061, 2001
Barker PE: Cancer biomarker validation: Standards and process. Ann N Y Acad Sci 983:142-150, 2003
Submitted March 20, 2005; accepted June 28, 2005.(Phyllis A. Gimotty, Patri)