|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 9 (March 20), 2007: pp. 1129-1134 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2006.08.1463 Identification of High-Risk Patients Among Those Diagnosed With Thin Cutaneous Melanomas
From The Melanoma Program of the Abramson Cancer Center, Center for Clinical Epidemiology and Biostatistics, Department of Biostatistics and Epidemiology, Department of Pathology and Laboratory Medicine, Department of Surgery, Department of Dermatology, and Department of Medicine at the University of Pennsylvania School of Medicine, Philadelphia, PA Address reprint requests to Phyllis A. Gimotty, PhD, Department of Biostatistics and Epidemiology, 631 Blockley Hall, 423 Guardian Dr, Philadelphia, PA 19104-6021; e-mail: pgimotty{at}cceb.med.upenn.edu
Purpose Most patients with melanoma have microscopically thin (  1 mm) primary lesions and are cured with excision. However, some develop metastatic disease that is often fatal. We evaluated established prognostic factors to develop classification schemes with better discrimination than current American Joint Committee on Cancer (AJCC) staging. Patients and Methods We studied patients with thin melanomas from the US population-based Surveillance, Epidemiology, and End Results (SEER) cancer registry (1988 to 2001; n = 26,291) and those seen by the University of Pennsylvania's Pigmented Lesion Group (PLG; 1972 to 2001; n = 2,389; Philadelphia, PA). AJCC prognostic factors were thickness, anatomic level, ulceration, site, sex, and age; PLG prognostic factors also included a set of biologically based candidate prognostic factors. Recursive partitioning was used to develop a SEER-based classification tree that was validated using PLG data. Next, a new PLG-based classification tree was developed using the expanded set of prognostic factors. Results The SEER-based classification tree identified additional criteria to explain survival heterogeneity among patients with thin, nonulcerated lesions; 10-year survival rates ranged from 89.1% to 99%. The new PLG-based tree identified groups using level, tumor cell mitotic rate, and sex. With survival rates from 83.4% to 100%, it had better discrimination. Conclusion Prognostication and related clinical decision making in the majority of patients with melanoma can be improved now using the validated, SEER-based classification. Tumor cell mitotic rate should be incorporated into the next iteration of AJCC staging.
Melanoma incidence rates continue to increase in the United States.1 Patients with American Joint Committee on Cancer (AJCC) stage I melanomas make up 78% of the cutaneous melanomas reported to National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) cancer registry. Of these, the majority (81%) have thin lesions (Breslow thickness of 1.00 mm).2 These patients generally have a good prognosis; however, some have a recurrence and eventually die of the disease. In SEER,2 the 5-year survival rate for these patients is 97.7%, a rate comparable to the 98% reported in Sweden3; the 10-year SEER survival rate is 95.1%, comparable to the 96.4% reported for such patients in both Australia and Germany.4,5 In SEER, 15% of melanoma deaths result from metastases of thin lesions. The current AJCC staging system6,7 uses two tumor-associated prognostic factors, ulceration and Clark's anatomic level, to dichotomize risk for patients with thin melanomas by defining a lower risk group, stage IA (level II/III and nonulcerated) and a higher risk group, stage IB (either level IV/V or ulcerated). We sought to develop for current use a classification scheme to identify high-risk patients with thin lesions for clinical trials, and to improve clinical decision making by using a classification tree-based methodology that would better stratify patients by risk of melanoma-specific death. We used data from the population-based SEER cancer registry to develop a SEER-based prognostic tree. This classification was then validated in a prospective cohort of similar patients evaluated and followed by the University of Pennsylvania's Pigmented Lesion Group (PLG; Philadelphia, PA). Four additional tumor characteristics were available for the PLG patients’ melanomas: growth phase, where tumors had either invasive radial growth phase (RGP) only or vertical growth phase (VGP); RGP regression; VGP mitotic rate; and VGP tumor-infiltrating lymphocytes (TILs).8,9 A new PLG-based prognostic tree was then developed based on these factors, in addition to those previously used to develop the AJCC stages, to investigate the significance of these characteristics as prognostic factors and to evaluate their candidacy for inclusion in the next revision of the AJCC staging system.
Study Populations In SEER,10 there were 26,291 patients in 11 sites (748 to 4,069 patients per site) identified between 1988 and 2001 who had thin primary cutaneous melanomas with no evidence of metastasis and complete data on the six prognostic factors included in AJCC staging11: thickness, level, ulceration, anatomic site, sex, and age (see Gimotty et al2 for additional details about these patients). Between 1972 and 2001, there were 2,389 patients with thin melanomas evaluated within 1 year of definitive therapy and subsequently followed by the PLG who met the study inclusion criteria used for the SEER patients.
Survival Status and Follow-Up
Prognostic Factors
Statistical Methods
Receiver operating characteristic (ROC) curves for predicting 10-year survival and the area under these curves (AUCs) were computed for each classification scheme using 1,076 PLG patients who had a melanoma-related death within 10 years of the definitive treatment for their primary site or had at least 10 years of follow-up. A 95% CI for the mean of the differences between the paired AUCs was obtained using 1,000 bootstrap samples.14 The survival and ROC curve analyses were done in SAS version 9.1 (SAS Institute, Cary, NC).15
Characteristics of Patients in SEER and PLG Cohorts As expected given the very large sample sizes, there were statistically significant differences between the cohorts (Fig 3B). However, the proportions in the PLG cohort associated with all but one of the risk factors (male patients, axial melanomas, lesions that were ulcerated, or level IV/V) were within the ranges observed within the 11 SEER regions (Fig 3A). Only the proportion of patients  60 years of age was significantly lower in the PLG cohort compared with the other SEER sites. The mean thickness of melanomas for SEER and PLG patients was 0.46 mm and 0.53 mm, respectively. Among the PLG patients, 58% had lesions with VGP. Of these 1,380 VGP lesions, 44% were mitogenic, 61% had TIL, and 16% had RGP regression.
Expanded AJCC Classification of Thin Melanomas To foster its immediate application, the first prognostic tree was developed with the prognostic factors used in the development of the current AJCC staging system: thickness, level, ulceration, anatomic site, sex, and age (Fig 1A). Of note was that the first and second factors selected for this classification tree by the algorithm were ulceration and level, respectively—the factors that define thin stage IA (T1a; nonulcerated level II/III) and IB (T1b; IV/V or ulcerated) melanomas in the AJCC staging system (Fig 1A, shaded tan). As we hypothesized, additional factors were then identified within these two groups, defining an expanded AJCC classification that characterized important differences in survival experiences. Thin stage IA melanomas were best characterized with six classes (classes 1 to 6; Fig 1B), the prognostic profiles of which were defined by thickness and then specific combinations of level, age, anatomic site, and sex. For example, based on the prognostic tree (Fig 1A) a male who presented with a 0.81-mm, nonulcerated melanoma (stage IA or T1a in the current AJCC staging system) would be among a class of patients with an estimated 10-year survival rate of 90.6% (class 6; Fig 1B). This rate is much lower than that expected for AJCC stage IA (97.4% in the SEER2 cohort and 97.2% in the PLG cohort) and is comparable to the 10-year survival rates of patients with stage IB melanomas (90.2% in the SEER2 and 92.6% in the PLG cohorts).
Validation of the Expanded Classification for Thin, Nonulcerated Melanomas
A New Classification of Thin Melanomas
Comparison of the Two Prognostic Trees
The ROC curves for 10-year survival using the two prognostic trees were estimated using 1,076 PLG patients with at least 10 years of follow-up (data not shown). The AUCs were 0.76 and 0.83 for the expanded AJCC tree and the new PLG prognostic tree, respectively. The AUC for the new prognostic tree was significantly higher than that for the expanded AJCC tree; the AUC mean difference was 0.06 (95% CI, 0.016 to 0.109).
This study had two objectives. First, we sought to re-evaluate the AJCC prognostic factors using SEER patients to develop for current clinical application an improved and easily accessible prognostic classification. Second, we evaluated four biologically relevant, readily ascertained candidate prognostic factors available for PLG patients (but unavailable in either the AJCC or the SEER cohorts) to identify potentially more discriminatory prognostic factors for consideration in the next version of the AJCC staging system. The current AJCC staging system6 that is used to identify patients with higher or lower risk of melanoma-related death was originally validated on a multi-institutional cohort of 17,600 patients that was not population based and wherein only 40% of patients had thin melanomas. The identification of prognostic factors for patients with thin melanomas was limited by the availability of only a few candidate variables recorded in this cohort, and by both the lack of long-term follow-up and the limited number of melanoma-specific deaths. Our investigation, the most extensive of its kind, included 26,291 patients with thin melanomas from a population-based registry with standard, AJCC-based prognostic factors that we could use to leverage the richer set of prognostic factors collected on the 2,389 patients from the PLG.
Our re-evaluation of the AJCC prognostic factors included both model development and model validation. We used a tree-based classification analysis to develop a prognostic tree to improve risk estimation. The resulting classification scheme was an expansion of the current AJCC risk groups for thin melanomas wherein stages IA and IB are distinguished by level and ulceration. The expanded classification further subclassified patients in stage 1A (T1a) using anatomic site, thickness, and age to define patient profiles associated with differential survival rates. Of note, and further supporting the generalizability of our findings, these factors were also included in a classification tree for thin melanomas developed in a European population,16 wherein age, level, and anatomic site in thinner ( The PLG-based classification tree, developed using the four additional candidate prognostic factors (growth phase, RGP regression, VGP mitogenicity, and TIL) available for PLG patients, included two factors that appeared in the expanded AJCC classification (level and sex) and one new prognostic factor, VGP mitogenicity. This new classification identified a group of men with nonulcerated, level III/IV mitogenic lesions with a 10-year survival rate lower than of any of the classes in the expanded AJCC classification. We note that, as with all new prognostic models, the predictive value of the new prognostic tree needs to be evaluated in other cohorts. Mitogenicity, a feature of the VGP,8,9,12 is important beyond those factors that are used at present to stage thin melanomas. It also reflects the biology of tumor progression. Based on lesion histology, rates of tumor cell mitoses, and the expression of Ki67 (an immunohistochemical marker of cell division), we have shown that tumor progression in primary melanomas may be characterized by three stages: a phase of rapid tumor cell proliferation in the epidermis (the in situ RGP), followed first by a phase of invasion into the dermis and decreased proliferation (the invasive RGP), and then by a phase in which higher proliferative rates resume in the dermis as the VGP ensues.9,17 Mitogenicity, scored in the dermis as either present or absent, or as a mitotic rate, has been validated widely and is readily ascertainable, transportable, and generalizable.18 It has been demonstrated in other studies to be an independent prognostic factor for disease-free survival and metastasis,19-21 including metastasis to sentinel lymph nodes (SLNs)22-25 for patients with thin melanomas. Our data support the incorporation of mitogenicity into the next version of AJCC staging for melanoma, as suggested in a recent international consensus statement.26 Patients with thin melanomas make up the majority of patients seen today. The two prognostic models presented here will help to establish evidence-based guidelines for their management (eg, staging with SLN biopsy and establishing a plan of follow-up). For example, these models have the potential to guide the use of SLN biopsy. Lymphatic mapping and SLN biopsies have been shown to identify a subset of patients without palpable regional nodal metastases who nevertheless have micrometastases detected on histologic examination of the node(s) and a significant risk of subsequent, melanoma-specific death. This procedure provides staging information and prophylaxis against more extensive regional nodal failure and may improve survival for certain patient subgroups.27 It has appreciable medical costs and is associated with some, usually mild, morbidities. Although current guidelines28 recommend SLN biopsy for patients with stage IB and II melanomas, there are no evidence-based recommendations for those with thin lesions. Given that there is no consensus on how to group patients with thin lesions by risk, it is unclear which patients with thin melanoma should be recommended to receive more intrusive management (eg, SLN biopsy and close follow-up) and which patients should simply be observed less frequently to detect additional primary lesions and the rare local recurrence or metastasis.28 Our data clearly identify groups of patients with very low risk who should not have SLN biopsy (eg, class 1 in Fig 1A and class A in Fig 2A). We9 and others have also used the presence of the invasive RGP without VGP to define a group of patients with minimal risk of metastasis. The prognostic trees presented increase the proportion of patients identified as having minimal risk. Twenty-two percent of PLG patients in class 1 (Fig 1A) and 20% of those in class A (Fig 2A) had VGP lesions. Nevertheless, their survival rates were statistically indistinguishable from that of patients with pure, invasive RGP lesions. The expanded AJCC classification that we have developed and validated can be used now to select patients for study who have a clinically significant chance of death from melanoma by 10 years. These patients are also at increased risk of having a positive SLN biopsy. Although the outcome we used in constructing these models was melanoma-specific death (invariably from metastatic disease), rather than regional nodal metastasis, this outcome is likely strongly associated with regional nodal involvement. Recent studies of patterns of care indicate that surgeons are offering SLN biopsies to patients with level IV or ulcerated lesions,29 as recommended by current guidelines.28 Our data suggest that in addition to patients with level IV or ulcerated primaries, the 2,272 patients (8.6%) in classes 4 and 6 (Fig 1B) who have at least a 5% chance of death from metastasis in 10 years would also be appropriate participants for a trial designed to test the yield of positive SLN biopsies in patients with thin primary lesions. This study demonstrates the heterogeneity of disease-specific survival among patients with thin stage I melanomas,30 and indicates that more of this variability can be characterized by classification schemes that use more information than is currently included in the AJCC staging system. When validated, the new prognostic tree (Fig 2A) will better discriminate those with a significant risk of metastasis. It may serve to inform clinical trials and decision making until prognostic models with new biomarkers are evolved and tested.
The authors indicated no potential conflicts of interest.
Conception and design: Phyllis A. Gimotty, Jeffrey Botbyl, DuPont Guerry Provision of study materials or patients: David E. Elder, Douglas L. Fraker, Rosalie Elenitsas, Michael E. Ming, Lynn Schuchter, Francis R. Spitz, Brian J. Czerniecki, DuPont Guerry Collection and assembly of data: Phyllis A. Gimotty, Jeffrey Botbyl, Rosalie Elenitsas, DuPont Guerry Data analysis and interpretation: Phyllis A. Gimotty, Douglas L. Fraker, Jeffrey Botbyl, Kimberly Sellers, DuPont Guerry Manuscript writing: Phyllis A. Gimotty, Douglas L. Fraker, DuPont Guerry Final approval of manuscript: Phyllis A. Gimotty, David E. Elder, Douglas L. Fraker, Jeffrey Botbyl, Kimberly Sellers, Rosalie Elenitsas, Michael E. Ming, Lynn Schuchter, Francis R. Spitz, Brian J. Czerniecki, DuPont Guerry
We thank the many patients who have been seen at the Pigmented Lesion Clinic, as well as the investigators (L.P. Bucky, MD, L.S. Callans, MD, B. Chang, MD, K.T. Flaherty, MD, A.C. Halpern, MD, R. Hamilton, MD, D.M. Hershock, MD, E.J. Kim, MD, D.D. Larossa, MD, S.R. Lessin, MD, D. Low, MD, P. Van Belle, PhD, and J.T. Wolfe, MD) and staff (C. Doyle, R. Holmes, N. Lowden, I. Matozzo, M. Price, M. Synnestvedt, J. Thompson, P. Wahl) for their contributions during the last three decades to the collection of research data on which this report is based. The authors also thank April Fritz, Manager, Data Quality, in the SEER Program of the Division of Cancer Control and Population Sciences of the National Cancer Institute for her help with the interpretation of the SEER data.
Supported in part by the Specialized Program of Research Excellence (SPORE) on Skin Cancer (Grant No. CA-093372). Presented in part at the 41st Annual Meeting of the American Society of Clinical Oncology, May 13-17, 2005, Orlando, FL, and at the Keystone Symposium on Advances in the Treatment and Understanding of Melanoma, January 19, 2006, Santa Fe, NM. Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. Jemal A, Devesa SS, Hartge P, et al: Recent trends in cutaneous melanoma incidence among whites in the United States. J Natl Cancer Inst 93:678-683, 2001 2. Gimotty PA, Botbyl J, Soong S, et al: A population-based validation of the AJCC melanoma staging system. J Clin Oncol 23:8065-8075, 2005 3. Lindholm C, Andersson R, Dufmats M, et al: Invasive cutaneous malignant melanoma in Sweden, 1990-1999: A prospective, population-based study of survival and prognostic factors. Cancer 101:2067-2078, 2004[CrossRef][Medline] 4. McKinnon JG, Yu XQ, McCarthy WH, et al: Prognosis for patients with thin cutaneous melanoma. Cancer 98:1223-1231, 2003[CrossRef][Medline] 5. Eigentler TK, Buettner PG, Leiter U, et al: Impact of ulceration in stages I to III cutaneous melanoma as staged by the American Joint Committee on Cancer Staging System: An analysis of the German Central Malignant Melanoma Registry. J Clin Oncol 22:4376-4383, 2004 6. Greene F, Page D, Fleming I, et al: (eds): AJCC Cancer Staging Handbook (ed 6). New York, NY, Springer, 2002 7. 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 8. Elder DE, Murphy GF: Melanocytic tumors of the skin. Washington, DC, Armed Forces Institute of Pathology, 1991 9. 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 10. National Cancer Institute: Surveillance, Epidemiology, and End Results (SEER) Program Public-Use Data (1973-2001). National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2004, based on the November 2003 submission. www.seer.cancer.gov 11. 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 12. Elder DE, Gimotty PA, Guerry D: Cutaneous melanoma: Estimating survival and recurrence risk based on histopathologic features. Dermatol Ther 18:369-385, 2005[CrossRef][Medline] 13. Segal MR: Regression trees for censored data. Biometrics 44:35-47, 1988 14. Efron B, Tibshirani TJ. An Introduction to the Bootstrap. New York, NY, Chapman & Hall, 1993 15. SAS Institute: SAS/STAT User's Guide, Version 9. Cary, NC, SAS Institute Inc, 2005 16. Leiter U, Buettner PG, Eigentler TK, et al: Prognostic factors of thin cutaneous melanoma: An analysis of the Central Malignant Melanoma Registry of the German Dermatological Society. J Clin Oncol 22:3660-3667, 2004 17. Gimotty PA, Van Belle P, Elder DE, et al: The biologic and prognostic significance of dermal Ki67 expression, mitoses and tumorigenicity in thin invasive cutaneous melanoma. J Clin Oncol 23:8048-8056, 2005 18. Scolyer RA, Shaw HM, Thompson JF, et al: Interobserver reproducibility of histopathologic prognostic factors in primary cutaneous melanomas. Am J Surg Pathol 27:1571-1576, 2003[CrossRef][Medline] 19. Massi D, Franchi A, Borgognoni, et al: Thin cutaneous malignant melanomas (< or =1.5 mm): identification of risk factors indicative of progression. Cancer 85:1067-1076, 1999[CrossRef][Medline] 20. 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[CrossRef][Medline] 21. 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 22. Bedrosian I, Faries MB, Guerry D, et al: Incidence of sentinel node metastasis in patients with thin primary melanoma ( 23. Oliveira-Filho RS, Ferreira LM, Biasi LI, et al: Vertical growth phase and positive sentinel node in thin melanoma. Braz J Med Biol Res 36:347-350, 2003[Medline] 24. 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 11:247-258, 2004 25. Kesmodel SB, Karakousis GC, Botbyl JD, et al: Mitotic rate as a predictor of sentinel lymph node positivity in patients with thin melanomas. Ann Surg Oncol 12:449-458, 2005 26. Atkins MB, Elder DE, Essner R, et al: Innovations and challenges in melanoma: Summary statement from the First Cambridge Conference. Clin Cancer Res 12:2291s-2296s, 2006 (suppl 7) 27. Morton DL, Thompson JF, Cochran AJ, et al: Sentinel-node biopsy or nodal observations in melanoma. N Engl J Med 355:1307-1317, 2006 28. Melanoma: Clinical practice guidelines in oncology. J Natl Compr Canc Netw 3:46-60, 2004 29. Cormier JN, Xing Y, Ding M, et al: Population-based assessment of surgical treatment trends for patients with melanoma in the era of sentinel lymph node biopsy. J Clin Oncol 23:6054-6062, 2005 30. Halpern AC, Marghob AA: Thin melanoma: Still "excellent prognosis" disease? J Clin Oncol 22:3651-3653, 2004 Submitted July 3, 2006; accepted December 8, 2006.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 test. L, level; U, ulceration; M, male; O, age > 60 years; S, axial site.
