Gene Expression Signatures for Tumor Progression, Tumor Subtype, and Tumor Thickness in Laser-Microdissected Melanoma Tissues

Human Cancer Biology

Gene Expression Signatures for Tumor Progression, Tumor Subtype, and Tumor Thickness in Laser-Microdissected Melanoma Tissues Jochen Jaeger,1Dirk Koczan,2 Hans-Juergen Thiesen,2 Saleh M. Ibrahim,2 Gerd Gross,3 Rainer Spang,1and Manfred Kunz3

Abstract Purpose:To better understand the molecular mechanisms of malignant melanoma progression and metastasis, gene expression profiling was done of primary melanomas and melanoma metastases. Experimental Design: Tumor cell ^ specific gene expression in 19 primary melanomas and 22 melanoma metastases was analyzed using oligonucleotide microarrays after laser-capture microdissection of melanoma cells. Statistical analysis was done by random permutation analysis and support vector machines. Microarray data were further validated by immunohistochemistry and immunoblotting. Results: Overall, 308 genes were identified that showed significant differential expression betweenprimary melanomas and melanoma metastases (false discovery rate V 0.05).Significantly overrepresented gene ontology categories in the list of 308 genes were cell cycle regulation, mitosis, cell communication, and cell adhesion. Overall, 47 genes showed up-regulation in metastases. These included Cdc6, Cdk1, septin 6, mitosin, kinesin family member 2C, osteopontin,and fibronectin. Down-regulated genes included E-cadherin, fibroblast growth factor binding protein, and desmocollin 1 and desmocollin 3, stratifin/14-3-3r, and the chemokine CCL27.Using support vector machine analysis of gene expression data, a performance of >85% correct classifications for primary melanomas and metastases was reached. Further analysis showed that subtypes of primary melanomas displayed characteristic gene expression patterns, as do thin tumors (V1.0 mm Breslow thickness) compared with intermediate and thick tumors (>2.0 mm Breslow thickness). Conclusions: Taken together, this large-scale gene expression study of malignant melanoma identified molecular signatures related to metastasis, melanoma subtypes, and tumor thickness. These findings not only provide deeper insights into the pathogenesis of melanoma progression but may also guide future research on innovative treatments.

The incidence of malignant melanoma is steadily increasing Breslow (2). It could be shown that tumors of a few-millimeter with a present lifetime risk of 1 in 75 among the Caucasian thickness already show a high potential for metastasis with a population (1). The underlying factors for this phenomenon fatal outcome for the patient. In the metastatic stage, melanoma are largely unknown. After diagnosis of malignant melanoma, patients have only few treatment options, consisting of the single most important factor for the prognosis of melanoma monochemotherapies with dacarbazine (DTIC) or temozolo- patients is vertical tumor thickness as described earlier by mide, or polychemotherapy regimens combining DTIC with other chemotherapeutic agents such as cisplatin and 1,3- bis(2-chloroethyl)-1-nitrosourea (3–5). Although significant Authors’ Affiliations: 1Department of Computational Molecular Biology, Max clinical response rates were achieved by these treatment Planck Institute for Molecular Genetics, Berlin, Germany and 2Institute of modalities, there was no substantial effect on the overall 3 Immunology and Department of Dermatology and Venereology, University of survival of these patients. Unfortunately, little is known about Rostock, Rostock, Germany Received 7/24/06; revised 10/5/06; accepted 11/28/06. factors that contribute to the process of melanoma progression Grant support: Deutsche Krebshilfe grant 70-2819 (M. Kunz) and National and metastasis. Genome Research Network grant 031U209(R. Spang). In recent years, DNA microarray technology has fostered The costs of publication of this article were defrayed in part by the payment of page hopes for a more complete understanding of the mechanisms charges. This article must therefore be hereby marked advertisement in accordance of tumor progression and metastasis in a variety of tumors, with 18 U.S.C. Section 1734 solely to indicate this fact. Note: Supplementary data for this article are available at Clinical Cancer Research including malignant melanoma (6–8). A series of studies done Online (http://clincancerres.aacrjournals.org/). on colon carcinoma, pancreatic adenocarcinoma, lung adeno- Requests for reprints: Manfred Kunz, Department of Dermatology and carcinoma, and breast carcinoma showed tumor-specific gene Venereology, University of Rostock, Augustenstr. 80-84, 18055 Rostock, expression compared with normal tissues (9–13). More Germany. Phone: 49-381-4949708; Fax: 49-381-4949702; E-mail: manfred. [email protected]. recently, a comprehensive microarray study on metastasis- F 2007 American Association for Cancer Research. related gene expression analyzed different primary tumors doi:10.1158/1078-0432.CCR-06-1820 from breast adenocarcinoma, prostate adenocarcinoma and

Clin Cancer Res 2007;13(3) February1,2007 806 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Molecular Signatures of Malignant Melanoma medulloblastoma, and metastases of the same tumor types (14). obtained from 41 different patients after surgical excision of tumors. A universal molecular signature for metastasis was suggested. The vertical tumor thickness of primary melanomas ranged from 0.38 Among the 17 signature genes were important mediators for the to 11.00 mm according to Breslow (2), including one in situ melanoma. interaction of tumor cells with the extracellular matrix, such as A complete list with detailed information about primary melanoma samples including thicknesses of biopsies is given in Supplementary type I collagen a1/a2. Table S1A.4 Immunohistochemistry was done on an independent set of A large-scale gene expression study based on the analysis of 20 primary melanomas and 20 melanoma metastases from either melanoma cell lines of different aggressiveness was presented archival paraffin-embedded or frozen material, depending on the by Clark et al. (15). RhoC, a member of the family of Rho antibodies used for immunostaining. The vertical tumor thickness of GTPases, was shown to be highly expressed in metastatic these primary melanomas ranged from 0.40 to 12.20 mm according to melanoma cells, when compared with their nonmetastatic Breslow. Metastatic lesions in both groups were cutaneous or s.c. counterparts of the same genetic background. The differences in metastases. The presented study was approved by the local Ethic gene expression between cell lines of both stages were most Committee at the University of Rostock, and informed consent for significant for genes involved in extracellular matrix and microarray analyses of tumors was obtained from all patients. cytoskeleton organization [e.g., fibronectin, collagen subunits Laser-capture microdissection of tumor tissue. Laser-capture micro- a a dissection was done as described by Bonner et al. (19). Cryoprepara- 2(I) and 1(III), matrix Gla protein, fibromodullin, biglycan, and A b tions of tumor specimens were cut into sections of 10- m thickness and thymosin 4]. put onto slides covered with a transparent plastic foil. Sections were To identify a metastatic phenotype in primary melanomas then stained using conventional H&E stain. Twenty to 30 sections of that might help to predict the clinical outcome of melanoma each tumor (primary tumors or metastases) were prepared. These were patients, Bittner et al. analyzed biopsies from primary subjected to laser-capture microdissection using a laser-capture micro- melanomas and melanoma metastases (16). Based on biosta- dissection Facility (P.A.L.M. Microlaser Technologies, Bernried, Ger- tistical and functional analyses, two major melanoma clusters many). Melanoma cells were microdissected under microscopic control were identified. Cluster I contained less aggressive melanoma by an experienced histopathologist. For the analysis of primary tissues, and cluster II contained highly aggressive melanoma melanomas, the entire thickness of each tumor was microdissected. tissues. Cluster I showed reduced expression for integrin h , This included intraepidermal melanoma cells, when these cells were 1 part of clearly demarcated tumor cell nests. Single cells in the epidermis integrin h , syndecan, vinculin, and fibronectin. This study 3 were not included. Metastases generally had a diameter of 6 to 8 mm further underlined the particular role of cytoskeletal and and were cut into halves. One half was used for laser microdissection, extracellular matrix molecules for melanoma progression. A and the other half was used for histopathology. In metastases, central and more recent study based on gene expression patterns of 58 peripheral areas were used for microdissection. Necrotic areas, when primary melanomas showed that gene expression patterns in present, were avoided. Microdissected tissues were captured in Eppendorf primary tumors may indeed help to predict the clinical tubes and processed for RNA extraction. outcome of melanoma patients (17). Gene expression patterns RNA extraction, labeling, and hybridization of microarray targets. in primary tumors differed between patients with a 4-year Total RNA was extracted from tumor tissues using commercially distant metastasis-free survival from those who developed available systems (RNeasy kit, Qiagen, Hilden, Germany). RNA probes A metastases within this time. A large series of molecules with were labeled according to the supplier’s instructions using 1 g of total RNA (Affymetrix, Santa Clara, CA). Analysis of gene expression was enhanced expression in the bad prognosis group belonged to carried out using the Human Genome U133A microarray (Affymetrix), the functional groups of cell cycle regulation, mitosis, and DNA which contains 22,215 probe sets (omitting control probe sets). replication, such as Cdc2, Cdc6, CENPF, and proliferating cell Hybridization and washing of gene chips were carried out according nuclear antigen. to the supplier’s instructions. A laser scanner (Gene Array Scanner from A comprehensive study on different stages of malignant Hewlett-Packard, Boeblingen, Germany) was used for reading out melanoma development was published recently, analyzing microarrays. specimens from benign melanocytic nevi, primary melanomas, Immunohistochemistry. Five-micrometer sections of paraffin- and melanoma metastases (18). A major finding of this study embedded or frozen tissues from 20 primary melanomas and 20 was the identification of two different gene patterns found in melanoma metastases were prepared for immunohistochemistry. To metastases reflecting those in vertical or radial growth phase improve antigen recognition in paraffin sections, a 1-min heat treatment was done in a pressure cooker, containing 1 liter of 10 cells of primary melanomas. These findings argue for a mmol/L sodium citrate buffer (pH 6.5). The following antibodies were metastatic gene pattern present in a subtype of primary tumors. used for immunostaining: anti–p34/cyclin-dependent kinase 1 (Cdk1; In the present report, the expression of 22,283 probe sets was DM114P, mouse monoclonal; Acris Antibodies, Hiddenhausen, Ger- analyzed in a series of laser-microdissected tissues from 41 many), anti-Cdc6 (CC30, mouse monoclonal; Calbiochem, Bad Soden, primary melanomas and melanoma metastases using oligonu- Germany), anti-desmocollin 1 (anti-DSC1; H44951, mouse monoclo- cleotide microarrays. Overall, 389 probe sets (representing 308 nal; Biodesign, Saco, ME), and anti-CCL27 (AF376, goat polyclonal; different genes) were identified that showed significant R&D Systems, Bad Nauheim, Germany). Immunodetection was done differential expression between both disease stages. A predictive with a commercially available system (ZytoChemPlus AP Broad diagnostic model [support vector machine (SVM)] for discrim- Spectrum Bulk kit, Zytomed, Berlin, Germany) using Permanent APRed inating primary tumors and metastases was trained, and a (Zytomed) as chromogen. Counterstaining was done with hematoxylin. To evaluate differential expression levels of homogeneously staining performance of >85% correct classifications was reached in Cdk1 and Cdc6 in primary melanomas and melanoma metastases, the cross-validation. following four-point scale scoring system was used: 0, negative staining; +1, weak staining; +2, moderate staining; +3, strong staining. To Materials and Methods

Tissue specimens. For microarray analyses, biopsy material of 19 4 Supplementary data are available at http://compdiag.molgen.mpg.de/supplements/ primary melanomas and 22 cutaneous melanoma metastases was Melanoma06.

www.aacrjournals.org 807 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology evaluate differential expression levels of DSC1 and CCL27, which diagnosis that shows the patient-specific confidence of molecular showed patchy staining or staining of isolated cells, the four-point scale diagnosis. This was achieved by running cross-validation 1,000 times scoring system was based on the percentage of positively staining cells with different random assignments to the 10-fold cross-validation bins in four different high power fields (400): 0, 0% to 5%; +1, 6% to of patients and recording how many times the SVM misclassified each 20%; +2, 21% to 50%, +3; 51% to 100% positively staining cells. sample. In 10-fold cross-validation, the data is divided into 10 equally Evaluation was done by two independent investigators. sized batches. Iteratively, one of the batches is set aside as a test set, and Immunoblots. The nonmetastasizing 1F6 melanoma cell line was a classifier is learned on the rest. This is done so that each datum was kindly provided by G.v. Muijen (Institute of Pathology, Nijmegen, The exactly once in a test set. The accuracy of the classifier is finally Netherlands; ref. 20). Melanoma cells were lysed on ice for 30 min determined by counting how many times the data in the test set was using radioimmunoprecipitation assay buffer. Forty micrograms of total assigned to the correct label. The SVM was trained using a linear kernel protein extract were denatured in electrophoresis sample buffer for and feature selection of the 100 genes with the highest absolute 5 min at 95jC and subjected to SDS-PAGE. Gels were electroblotted onto significance analysis of microarrays scores. To find novel molecular nitrocellulose membranes (Highbond ECL, Amersham, Braunschweig, subentities of melanomas and metastases in our samples, an Germany) and subjected to immunodetection. The following primary unsupervised class discovery analysis using the ISIS algorithm program antibodies were used: anti-DSC1 (H44951, mouse monoclonal; was done (28). The ISIS algorithm searches for binary class partitions Biodesign), anti-CCL27 (AF376, goat polyclonal; R&D Systems), anti- that are clearly separated based on a subset of genes. Due to memory cytokeratin 10 (CK10; DM055, mouse monoclonal; Acris Antibodies), and time constraints, we pre-filtered the genes to the 9,000 highest anti-CK14 (DM3066, mouse monoclonal; Acris Antibodies), anti– variance, expressed genes and set the default setting for the number of stratifin/14-3-3j (ab14123, mouse monoclonal; Abcam, Cambridge, possible candidate splits to 3,000. Finally, the splits were further refined United Kingdom), anti–fibroblast growth factor binding protein (anti- by only keeping splits that had >60 (0.5%) differentially regulated V FGFBP; MAB1593, mouse monoclonal; R&D Systems), anti-integrin h4 genes at a FDR 0.05. For statistical analysis of immunohistochemical 2 (ab6136, mouse monoclonal; Abcam), anti-S100A2 (S6797, mouse staining patterns, a m test according to Pearson was done. Differences monoclonal; Sigma-Aldrich, Munich, Germany), anti-osteopontin in staining intensities of tissue sections with P V0.05 were regarded as (BAF1433, biotinylated goat polyclonal; R&D Systems). Signal detec- statistically significant. tion was done by appropriate anti-mouse or anti-goat horseradish peroxidase–coupled secondary antibodies (purchased from Becton Dickinson, Heidelberg, Germany), or by horseradish peroxidase– Results coupled streptavidin (S2438, Sigma-Aldrich). A standard enhanced Molecular signatures of melanoma metastases. Oligonucleo- chemiluminescence reaction (Amersham) was done for signal visual- ization. tide microarrays harboring >22,000 probe sets were used to Pre-processing of microarray data. For pre-processing of microarray analyze gene expression patterns of 19 primary melanomas and data, a background correction, normalization on probe level, and probe 22 melanoma metastases after laser-capture microdissection set summarization were done. The background correction was done of tumor cells. By this means, 308 differentially expressed similarly to Microarray Suite 5.0 (Affymetrix, 2001), but negative values genes were identified with an estimated FDR V0.05 (Fig. 1A). were not truncated. Probe level normalization was done using the Interestingly, the vast majority of 261 genes showed reduced variance stabilization method by Huber et al. (21). Finally, probe set expression in metastases. However, 47 genes showed up- summarization was done using a median polish fit of an additive regulation in metastases. Among the up-regulated genes were model (22). Cdc6, Cdc28/Cdk1 subunit 2 (CKS2), septin 6 (SPT6), mitosin, Statistical analysis of microarray data and immunohistochemical kinesin family member 2C (KIF2C), and serine/threonine kinase 6 staining patterns. To find genes with statistically significant differences in gene expression between different clinical phenotypes, genes were (STK6). This set of genes also included well-known tumor ranked according to a regularized t score (23). False discovery rates progression factors, like osteopontin (SPP1), fibronectin (FN1), (FDR) for the lists of top ranking genes were calculated based on 10,000 and members of the MAGE family of differentiation antigens, random permutations of the class labels (24). Differences in gene and less-known tumor progression molecules, such as thrombo- expression were regarded as significant when the FDR of the resulting spondin 4, biglycan, and NCAM1. The complete list of up- lists did not exceed 0.05. The expression levels of these genes were regulated genes, including information about their main visualized in color-coded heat maps. In these plots, genes were hierar- functions, is given in Supplementary Table S2. chically clustered using complete linkage. The samples were grouped Molecules that showed down-modulation in metastases were by phenotype, and each group is separately hierarchically clustered E-cadherin (CDH1), FGFBP, keratin 10 (KRT10), DSC1 and for better visual comparability. Using the R Software for Statistical DSC3, desmoglein 1 (DSG1), stratifin/14-3-3j, chemokine Computing and the Bioconductor package GOstats, every gene list was further examined for significant overrepresentation of biological CCL27, tumor protein p73-like (TP73L), p21-activated kinase 6 processes and pathways defined by gene ontology (GO) categories. To (PAK6), and FGF receptors 1 and 2 (FGFR2 and FGFR3). The derive diagnostic signatures, SVM (25, 26), combined with a regularized complete list of down-regulated genes, including information t score–based feature selection filter, were used. To compensate for about their main functions, is given in Supplementary Table S3. unbalanced group sizes, we adjusted the class weights within the SVM Some of these genes had been described in earlier studies as according to the group sizes. Predictive performance was assessed using being down-regulated during tumor progression in malignant the MCRestimate package (27). Using this package, the optimal number melanoma (e.g., E-cadherin; ref. 29). However, others have not of genes and the optimal variable setting of the SVM were determined yet been described to be expressed by melanoma tumors (e.g., in a nested cross-validation setting. Cross-validation was repeated FGFBP, KRT10, DSC1, DSC3, stratifin/14-3-3j, and CCL27). 10 times with a 10-fold outer and 10-fold inner loop. The outer loop Using Fisher’s exact test, the list of 308 differentially expressed construction ensures an unbiased estimation of predictive performance, even when using optimal variables. Predictive performances are genes was further examined for significant overrepresentation of generally contrasted with prevalence. The prevalence is the number of biological processes as defined by GO categories. Genes with the samples in the larger of the two group divided by the total number of GO terms cell cycle (GO:7049), mitotic cell cycle (GO:0278), M samples. Thus, it is the fraction of correct classifications when assigning phase of mitotic cell cycle, (GO:0279), mitosis (GO:7067), and all samples to the bigger group. We also implemented an in silico panel chromosome condensation (GO:30261) showed statistically

Clin Cancer Res 2007;13(3) February1,2007 808 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Molecular Signatures of Malignant Melanoma

Fig. 1. Molecular differentiation based on gene expression profiles between primary melanomas and melanoma metastases, different melanoma subtypes, and primary melanomas of different tumor thickness. A, A color-coded heat map was generated from differentially expressed genes after comparing gene expression in primary melanomas and melanoma metastases using 10,000 permutations. Only genes were included with a FDR V0.05. Overall, 308 differentially expressed genes were identified, which clearly separated both tumor stages. Increased (yellow) and decreased (blue) gene expression referring to the mean of each individual gene over all patients. Right margin, representative genes. B, the same analysis was done as described in (A) for the identification of differentially expressed genes in different melanoma subtypes (NM versus SSM). Overall, 60 differentially expressed genes were identified. The one ALM tested in this study showed a gene expression pattern similar to that of NM. C, the same analysis was done as described in (A)and(B) for primary melanomas of different tumor thickness (V1.0 mm Breslow thickness versus >2.0 mm Breslow thickness). Overall, 199 genes showed differential gene expression. The majority of these (172 genes) were up-regulated in thin (V1mm Breslow thickness) melanomas, and only a minority (27 genes) was up-regulated in intermediate and thick (>2 mm Breslow thickness) melanomas.

significant enrichment in the list of up-regulated genes in TP73L. Some of these genes (e.g., connexin 43 and DSG1) had metastases. Genes with the GO terms epidermis development been shown to be down-regulated during tumor progression (GO:8544), ectoderm development (GO:7398), cell communi- and advanced tissue invasion of melanomas (29). Among those cation (GO:7154), cell-cell adhesion (GO:16337), and homo- up-regulated in NM were matrix metalloproteinase 16 (MMP16), philic cell adhesion (GO:7156) showed statistical significant BCL2-related protein A (BCL2A1), intercellular adhesion molecule enrichment in the list of down-regulated genes in metastases. 1 (ICAM1), and carcinoembryonic antigen–related cell adhesion Figure 2 shows the hierarchical structure of these GO terms. molecule 1 (CEACAM1), representing molecules involved in Significantly enriched nodes/biological processes are indicated. tissue invasion and cell-cell adhesion. In accordance with our Taken together, primary melanomas and metastases show findings, enhanced ICAM1 and CEACAM1 expression has been significant differential gene expression. Functional categories of described earlier in advanced melanomas (30). differentially expressed genes argue for a particular role of Based on currently available information, the mean tumor cellular proliferation, cell cycle regulation, cell adhesion, and thickness of primary melanomas at the time of first diagnosis is cell-extracellular matrix interaction as central mechanisms for 0.3 to 0.5 mm (31, 32). In the presented analysis, the group of tumor progression in melanoma. SSM included two SSM (SSM6 and SSM13) with unusual high Molecular signatures of melanoma subtypes. Our next anal- tumor thickness of 5.67 and 5.4 mm, respectively. Interestingly, yses addressed the question of whether melanoma subtypes these two thick SSM showed a gene pattern similar to that of may be differentiated by gene expression patterns. When SSM and not of NM, which further supports the notion that comparing gene expression profiles of SSM and NM, a series there is indeed a clinical and molecular difference between of 67 probe sets (60 genes) was identified with a FDR V0.05 both melanoma subtypes. In summary, NM and SSM show (Fig. 1B). Genes with the following GO terms showed different gene expression patterns. Functionally, SSM show statistically significant enrichment in the list of up-regulated enhanced expression of genes involved in cell-cell contact and genes in SSM: cell adhesion (GO:7155), ectoderm development cell-cell communication. (GO:7398), morphogenesis (GO:9653), and intercellular junc- Molecular signature of tumor thickness in primary melanomas. tion (GO:45216). When addressing particular genes, SSM The most important prognostic factor for malignant melanoma displayed enhanced expression for FGFR2 and FGFR3, connexin is vertical tumor thickness, also termed Breslow thickness. 43 (GJA1), annexin A8 (ANXA8), small cell lung carcinoma To address the question, whether tumor thickness of primary cluster (CD24), chemokine ligand 14 (CXCL14), DSG1, and tumors is reflected by a particular gene expression pattern,

www.aacrjournals.org 809 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

Fig. 2. GO tree of significantly enriched biological processes in differentially expressed genes between primary melanomas and metastases. GO terms were determined for all significantly differentially expressed genes in the comparison of primary melanomas and metastases. Nodes (GO terms) with upward pointing arrows indicate enriched GO terms in the list of up-regulated genes in metastases. Nodes with downward pointing arrows indicate enriched GO terms in the list of down-regulated genes in metastases. Levels of significance are indicated by arrowheads (single arrowhead, P < 0.1; double arrowhead, P < 0.01; triple arrowhead, P < 0.001). Only statistically significant nodes and their parental nodes are displayed.

thin primary tumors (V1.0 mm Breslow thickness) were Misclassified samples were NM1, NM2, NM3, Meta1, and compared with intermediate and thick primary tumors (>2.0 Meta18. For a couple of samples (Meta2, Meta19, ALM1, and mm Breslow thickness). These cutoff points were chosen based SSM13), the SVM diagnosis results were ambiguous between on an earlier study correlating tumor thickness with the different cross validation runs, indicating that for these patients patients’ prognosis in 5,000 melanoma patients (33). When the expression profiles do not contain sufficient information to comparing both groups, a list of 240 probe sets (representing decide whether patients had primary tumors or metastases. In 199 genes) was identified which showed differential expression these analyses, the accuracy for distinguishing SSM and NM was with an estimated FDR V0.05 (Fig. 1C). Interestingly, 116 89% (with a prevalence of 72%) using 15 genes. The accuracy (58%) of these 199 genes overlapped with differentially for distinguishing SSM and Meta was 91% (with a prevalence of expressed genes of primary melanomas and metastases. 63%) using 50 genes. NM could not be distinguished from Meta Genes with the GO terms cell communication (GO:7154), with an accuracy higher than prevalence. cell adhesion (GO:7155), and ectoderm development Taken together, these analyses showed that by use of a (GO:7398) were significantly enriched in the list of genes with supervised classification method, a high level of prediction enhanced expression in thin melanomas. These categories included DSG1, connexin 43 (GJA1), epidermal growth factor receptor (EGFR), ANXA8, and CD24. Among the up-regulated genes in intermediate and thick melanomas were BCL2A1, cadherin 19 (CDH19), BH-protocadherin (PCDH7), regulator of G-protein signaling 20 (RGS20), thymosin-like 8 (TMSL8), catenin a-like 1 (CTNNAL1), and Ets variant gene 1 (ETV1). Overall, genes with the GO terms regulation of cellular process (GO:50794), regulation of signal transduction (GO:9966), and cell communication (GO:7154) were significantly enriched in the list of genes with enhanced expression in intermediate and thick melanomas. Taken together, these findings argue for the development of a metastatic signature in primary tumors with increasing tumor thickness, which supports the concept of tumor thickness as a major prognostic factor. Molecular classification by SVM. We then addressed the question, whether gene expression profiles may accurately predict the tumor stage (primary melanoma or metastasis) of Fig. 3. Class prediction of primary melanomas and metastases using SVM analysis of gene expression profiles.The percentage of misclassification is depicted for each our samples and calculated the estimated accuracy of this tumor sample (primary melanoma or metastasis) when repeating 1,000 random prediction. For this purpose, SVM were applied (34). In an 10-fold cross-validations with a linear kernel SVM. The plot is sorted by misclassification rate and first shows all metastasis samples and then all melanoma unbiased validation, using nested cross-validation, our predictor samples. Light gray columns, correct classification; dark gray columns, reached an accuracy of 85% correct classifications (with a misclassification. Positive values indicate how often a sample was classified as prevalence of 54%). Classifiers of this accuracy could be melanoma, and negative values how often it was classified as metastasis. Misclassifications were observed for NM1, NM2, and NM3, Meta1, and Meta18. designed with as little as 30 genes. Mainly NM were misclassified The oneALM was more often misclassified as metastasis than correctly classified as (Fig. 3), and most of them were misclassified in all runs. primary melanoma.

Clin Cancer Res 2007;13(3) February1,2007 810 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Molecular Signatures of Malignant Melanoma accuracy for tumor stages could be achieved, which further covery supported our findings of a molecular differentiation argues for particular biological stages represented by different not only between primary tumors and metastases but also melanoma subtypes or primary melanomas and metastases. between subtypes of NM. Moreover, unsupervised clustering Unsupervised class discovery. To find novel molecular sub- provided evidence for a molecular differentiation between thin entities of melanomas or metastases, we did an unsupervised and thick SSM. class discovery analysis using the ISIS algorithm (28). The ISIS Validation of microarray data by immunohistochemistry and algorithm finds bipartitions in a set of samples that show a immunoblotting. Immunohistochemistry of different melano- clear separation regarding the expression of a subset of genes. ma tissues was done to validate microarray data by an By use of this technique, a total of 20 bipartitions (splits) were independent method. A detailed list of the specimens and produced (Supplementary Fig. S1). Examples of differentially staining patterns is given in Supplementary Table S4. Figure 4 expressed genes of major splits (splits 9, 11, 15, and 19), and shows representative tissue sections from 20 primary melano- tumor samples forming clusters within these splits are shown in mas and 20 metastases immunostained for Cdc6, Cdk1, DSC1, Table 1. Gene patterns in split 15 were able to differentiate and CCL27/CTACK. These genes have been chosen because between primary melanomas and metastases. In accordance they represent new interesting molecules for melanoma with the above described findings, three NM (NM1, NM2, and pathogenesis. Cdc6 and DSC1 had not been described in NM3) grouped together with the majority of metastases. melanoma cells before, and Cdk1 and CCL27 had not been Interestingly, tumor thickness alone was not sufficient to shown to be differentially expressed in metastases compared attribute primary tumors to the metastasis cluster because the with primary melanomas. All four genes are involved in central two thick SSM (SSM6 and SSM13) did not group together with mechanisms that contribute to melanoma pathogenesis [i.e., NM1-3 and metastases in cluster 1. deregulated cell cycle control (Cdk1 and Cdc6), loss of cell These analyses also provided evidence that SSM might be adhesion (DSC1), and high immunogenicity of primary divided into two subentities. Split 11 showed a tumor cluster tumors (CCL27/CTACK)]. In microarray experiments Cdc6 consisting of SSM1 and SSM7 to SSM9, all primary tumors with and Cdk1 showed up-regulation, DSC1, and CCL27/CTACK low tumor thickness (i.e., 0.45 mm, 0.6 mm, in situ melanoma, showed down-regulation in metastases compared with primary and 0.95 mm, respectively). Top genes within split 11 were melanomas. Both Cdc6 and Cdk1 showed strongly positive lectin domaine 3B (CLEC3B), clusterin, dermatopontin,and immunohistochemical staining of metastases and moderate to COLA1, functionally all related to the cell-extracellular matrix weak staining of primary melanomas. Differences in staining interaction. The four thin primary SSM were also part of a intensities between metastases and primary melanomas were cluster in split 9. Split 19 showed a cluster of NM1 to NM3 and statistically significant with P < 0.001, as determined by m2 test. several metastases, which differed in gene expression patterns Interestingly, both molecules showed mainly cytoplasmic from all SSM, further underlining the molecular differences in expression in melanoma cells. This unexpected observation melanoma subtypes. Taken together, unsupervised class dis- was supported by earlier reports, which showed that Cdc6 is

Table 1. List of differentially regulated genes and tumor samples within splits 9, 11, 15, and 19

Split no. Major differentially Tumor cluster 1 Tumor cluster 2 Dependency expressed genes 9 ", FLJ12895; ", RNASE4; ", GAS1; #, SSM1, SSM2, SSM4, SSM7-9, SSM3, SSM5, SSM6, SSM10, Thickness MAGEA3; #, MAGEA6; #, MMP1; #, STK6 SSM12, SSM13 SSM11 NM1, NM2 NM3, NM4, NM5 ALM1 — Meta3, Meta4, Meta8, Meta11, Meta1, Meta2, Meta5, Meta6, Meta12, Meta17, Meta20 Meta7, Meta9, Meta10, Meta13-16, Meta18, Meta19, Meta21, Meta22 11 #, MBP; #, SERPINE2; ", CLEC3B; ", SSM1, SSM7-9 SSM2, SSM3-6, SSM10-13 Thickness CLU; ", DPT; ", OSR2; ", — NM1-5 COL1A1; ", PECAM1; ", TNXB — ALM1 Meta11 Meta1-10, Meta12-22 15 #, LGALS7; #, TACSTD2; #, KRT10; #, — SSM1-13 Metastasis KRT14; #, SFN; #, FGFBP1; #, NM1-3 NM4, NM5 DSC1; #, AQP3 ALM1 — Meta2-17, Meta19-22 Meta1, Meta18 19 ", NELL1; ", SCRG1; ", SPRY2; ", — SSM1-13 Subtype KHDRBS3; ", CTNNAL1; ", NM1-3 NM4, NM5 MMP16; ", AKT3; ", CDH2 ALM1 Meta5, Meta8, Meta9, Meta1-4, Meta6, Meta7, Meta12, Meta14, Meta20 Meta10, Meta11, Meta13, Meta15-19, Meta21, Meta22

NOTE: Binary splits were generated by unsupervised clustering. The complete list of splits is provided in Supplementary Fig. S1. ", up-regulated gene expression in cluster 1 compared with cluster 2. #, down-regulated gene expression in cluster 1 compared with cluster 2.

www.aacrjournals.org 811 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

Fig. 4. Validation of gene expression data by immunohistochemical staining of primary melanomas and melanoma metastases. Immunohistochemical staining of 20 primary melanomas (PM)and20 melanoma metastases (Meta)wasdone for Cdc6, Cdk1, DSC1, and CCL27. Representative core samples stained with either antibody. Corresponding genes of Cdc6 and Cdk1showed up-regulation in metastases; corresponding genes of DSC1 and CCL27 showed down-regulation in metastases compared with primary melanomas. Original magnification, 10 0 and 400 (inset).

nuclear in G1 phase of cell cycle but translocates to cytoplasm (SSM versus NM) and between tumors of different tumor after activation and phosphorylation and is mainly cytoplasmic thicknesses (tumors V1.0 mm compared with tumors >2.0 mm in S phase of cell cycle (35, 36). Cytoplasmic staining of Cdk1 tumor thickness). has also been reported in advanced lesions and metastases of Using different statistical approaches, it could be shown that esophageal adenocarcinoma (37). primary tumors were clearly distinguishable from metastases by In contrast to Cdc6 and Cdk1, DSC1 and CCL27/CTACK their molecular gene patterns. These findings argue that specific showed a more focal staining of tumor areas or isolated cell biological processes are active at different tumor stages. Major clusters. Both DSC1 and CCL27/CTACK showed higher biological functions as derived from GO categories involved in percentages of positively staining cells in primary melanomas this process were cell proliferation, cell cycle regulation, and compared with metastases. These differences were statistically mitosis, represented by the majority of up-regulated genes in significant with P < 0.010 for DSC1 and P < 0.001 for CCL27. metastases. These findings are in accordance with a large body Interestingly, DSC1 had not been described to be expressed in of data implicating that cancer development and progression melanoma cells before. Figure 4 shows representative samples are based on deregulated cellular proliferation and cell cycle immunostained for both molecules. control (38, 39). Indeed, in a variety of tumors activating Immunohistochemical analyses were extended by immunoblots for DSC1, CCL27, and a further series of molecules identified in the presented microarray study, which had not yet been described in melanoma cells (FGFBP, CK10, CK14, stratifin/14-3-3j, and integrin h4). As shown in Fig. 5, protein extracts of 1F6 melanoma cells showed moderate to strong expression of these proteins. Immunoblots for osteopontin and S100A2 were included as controls. The latter are known to be expressed by melanoma cells in vitro and in vivo and have also been detected in our microarray analyses. Taken together, these experiments further supported the presented microarray data, which identified new molecules not yet described in melanoma cells.

Discussion

A large-scale gene expression study is presented analyzing primary melanomas and melanoma metastases. To focus on gene expression in melanoma cells, tumor cells were excised from tissue biopsies using laser-capture microdissection. A major aim of the study was the identification of genes with enhanced expression in metastatic lesions, which might be of interest as targets for future innovative treatment approaches. Based on the analysis of >22,000 probe sets, 389 probe sets (representing 308 different genes) were shown to be differen- Fig. 5. Immunoblots of new melanoma cell proteins.Whole-cell lysates were prepared from 1F6melanoma cells, and the expression of indicated proteins was tially expressed between primary melanomas and metastases. analyzed by immunoblotting.The expression of these by melanoma cells is shown in Of these, a majority was down-regulated in the metastatic stage. the presented microarray analyses, but the majority of these had not been described However, 57 probe sets (representing 47 different genes) were in melanoma cells before. Osteopontin and S100A2 served as positive controls; both are known to be expressed by melanoma cells in vivo and in vitro. To control up-regulated in the metastatic stage. Gene expression profiles equal loading of proteins, blots were reprobed for h-tubulin expression. One were also able to differentiate between melanoma subtypes representative experiment of three independent experiments.

Clin Cancer Res 2007;13(3) February1,2007 812 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Molecular Signatures of Malignant Melanoma mutations in oncogenes, such as phosphatase and tensin model for melanoma progression based on a tightly regulated homologue deleted on chromosome 10, Akt kinase, and Ras, up-regulation and down-regulation of molecules involved inducing cellular proliferation have been identified. However, in cell-cell adhesion and cell-extracellular matrix interaction such mutations have rarely been detected in human melano- (29, 48). It was shown that benign melanocytes express high mas and melanoma metastases (1, 40). There is a considerable levels of E-cadherin, which mediates contact to epidermal body of evidence from experimental mouse melanoma models keratinocytes via E-cadherin-connexin 43 interactions. During that activating Ras mutations in combination with an transition from benign melanocytes to melanoma cells, inactivated INK4a/ARF (CDKN2a) tumor suppressor locus E-cadherin is down-regulated, and N-cadherin is up-regulated. induce spontaneous melanomas with consecutive distant In accordance with this, it could be shown in the presented metastases in mice (40–42). These findings argue for a partic- study that both E-cadherin and connexin 43 were down- ular role of deregulated cell cycle control in the pathogenesis regulated in NM compared with SSM. A series of further of malignant melanoma. However, little is known about the molecules (e.g., tenascin, ICAM-1, Mel-CAM) and members of molecules and pathways involved in these processes. In the the integrin family (e.g., avh3, a2h1, and a4h1 integrins) presented study, a series of candidate molecules were identified. involved in cell adhesion had been described to play a role in Among these were Cdc6, Cdk1, mitosin (CENPF), SPT6, and tumor progression and tissue invasion in melanoma (29, 49). KIF2C. In line with this, enhanced expression of Cdc6, Cdk1 Cell adhesion molecules also formed a major group of down- (CKS2), CENPF, and KIF2C in primary melanomas had been regulated molecules in thick tumors in the mentioned gene shown to be associated with bad prognosis, referring to a 4-year expression study of primary melanomas (17). distant metastasis-free survival of investigated patients (17). Interestingly, some of the overexpressed molecules in Enhanced gene expression of Cdc6 and Cdk1 in metastases in primary melanomas compared with metastases identified in our study was paralleled by overexpression of corresponding this study had not been described to be expressed in melanoma proteins, as shown by immunohistochemistry in an indepen- cells before (e.g., FGFBP, CK10, CK14, DSC1, DSC3, stratifin/ dent set of melanoma tumors. Cdc6 is an essential factor for 14-3-3j, and CCL27) but had been shown to be expressed by DNA replication regulated by E2F in mammalian cells. It had epidermal keratinocytes. Therefore, we did immunohistochem- been also described as a marker molecule for progressive istry and immunoblotting to further validate our findings. By cervical cancer and showed enhanced expression in a subset of this means, it could be shown that melanoma cells in in vivo non–small cell lung carcinomas compared with normal tissue and in vitro express the mentioned adhesion and cytoskeletal (43, 44). The Cdk complex cdc28-Cln is of central importance molecules as well as the CCL27 chemokine. In summary, the for cell division in yeast. The yeast cdc28 molecule corresponds presented data together with those of other groups underline to Cdk1 in the mammalian system and is a major regulator in the concept of local tumor progression and distant metastasis G2-M phase of the cell cycle. Here, we show that melanoma in melanoma based on specific losses and gains of adhesion metastases express high levels of Cdk1 mRNA and protein molecules and molecules facilitating migration and tissue compared with primary melanomas. These findings may open invasion. interesting perspectives for future melanoma treatment An important finding of the presented study was the approaches because a series of new small-molecule inhibitors molecular differentiation of tumors of different tumor thick- targeting cyclins and Cdks have already been, or will be, used nesses. We chose melanomas of less than 1 mm tumor in clinical trials in the near future (45). Among these are in thickness and compared these with melanomas thicker than particular inhibitors of Cdk1, such as flavopiridol and 2 mm. These cutoff points had been shown earlier to be of roscovitine. These molecules have been tested in chronic relevance for the prognosis of melanoma patients (33). Both lymphocytic leukemia, non-Hodgkin’s lymphoma, and kidney tumor stages could clearly be differentiated from each other cancer with remarkable response rates. Mitosin, also termed based on their gene expression profiles. The application of centromere protein F, is a multifunctional protein for cell unsupervised clustering methods even allowed further subclas- proliferation and is of importance for the function of the sification of a group of very thin SSM. Interestingly, there were a mitotic spindle checkpoint (46). It is active in the G2-M phase considerable number of genes overlapping between the of the cell cycle. Interestingly, mitosin has been shown to be an metastasis signature and the signature for tumor thickness. In independent predictor of recurrence in breast carcinomas (47). fact, 58% of differentially expressed genes in intermediate and A second major functional group identified, when comparing thick versus thin melanomas overlapped with those differen- gene expression in primary melanomas and metastases, tially expressed genes comparing metastases with primary included molecules involved in cell-cell adhesion and cell- melanomas. These findings are highly suggestive for a extracellular matrix interaction. This group comprised up- progressive change in gene expression patterns during tumor regulated and down-regulated genes in metastases. Prominent progression in melanoma. Moreover, these molecular data up-regulated genes involved in these processes were osteopontin, support the current concept of tumor thickness of primary fibronectin, biglycan, and thrombospondin 4. Osteopontin, fibro- tumors as the major prognostic factor for melanoma patients nectin, and biglycan have been described as progression factors (2, 5, 33). These overlapping genes might even be regarded as a for melanoma in recently published large-scale microarray metastatic signature in primary melanomas as described for studies from other groups analyzing either different melanoma other tumor entities (13, 14). tissues (16, 18) or melanoma cell lines of different aggressive- Among the genes that showed up-regulation in gene ness (15). Molecules with down-regulated expression in expression in intermediate and thick melanomas were BCL2A1, metastases were E-cadherin, FGFBP, CK10, DSC1, DSC3, different cadherins,andthymosin b.Interestingly,BCL2, stratifin/14-3-3j, and CCL27. These findings were partly cadherins, thymosin h, and Rho molecules had been shown overlapping with studies of Herlyn et al., who proposed a to contribute to melanoma progression in a series of

www.aacrjournals.org 813 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology independent studies (15, 29, 50). In a further recently BCL2A1, ICAM1, and CEACAM1, emphasizing the role of published gene expression study, vertical growth phase of molecules supporting tissue invasion, proliferation, and primary melanomas was compared with horizontal growth adhesion mechanisms for tumor progression in NM. In phase (18). In accordance with our findings, down-modulated support of these findings, ICAM1 and CEACAM1 had been genes involved functional categories of cell adhesion and shown earlier to contribute to progressive tumor growth in extracellular matrix molecules, such as DSC2, MMP10, CDH3, melanomas (30). and integrin a2. Taken together, the presented data show that melanoma Based on gene expression profiles identified in our study, metastasis represents a specific biological stage of tumor different subtypes of melanomas could also be distinguished. progression with a particular gene pattern. A majority of up- Interestingly, two SSM with tumor thicknesses comparable with regulated genes in metastases fit the current pathogenic that of the NM showed gene pattern of SSM and not of NM. concepts of tumor progression and may serve as targets for This argues against a differentiation between both subtypes innovative treatment approaches. Moreover, we were able to merely based on different tumor thicknesses. These findings are show that melanomas of different thicknesses and different in accordance with epidemiologic data summarized in a recent melanoma subtypes are represented by particular gene expres- publication on prognostics factors of melanoma, showing that sion patterns. Further studies should be initiated to analyze risk profiles of both tumor subtypes remain different even after whether gene expression patterns in primary melanomas may correction for tumor thickness (51). predict the prognosis of patients, and whether gene expression The differences between the two major subtypes of patterns in metastases may be used for treatment monitoring in melanomas (SSM and NM) involved genes of cell adhesion, clinical trials. cell communication, epidermis development, and morphogenesis. SSM compared with NM showed enhanced expression of FGFR2 and FGFR3, connexin 43, ANXA8, and DSG1. Some Acknowledgments of these genes (e.g., FGFR2, connexin 43, and DSG1) had been We thank R.Waterstradt and H. Bergmann for excellent technical assistance and shown to play a role in melanoma development (28). Among Steinbeis Tranfer Center (Proteome Analysis Rostock) for providing the P.A.L.M. those with enhanced expression in NM were MMP16, laser-capture microdissection facility.

References 1. Tucker MA, Goldstein AM. Melanoma etiology: where 15. Clark EA, GolubTR, Lander ES, Hynes RO. Genomic 27. Ruschhaupt M, Huber W, Poustka A, Mansmann U. are we? Oncogene 2003;22:3042 ^ 52. analysis of metastasis reveals an essential role for A compendium to ensure computational reproducibil- 2. Breslow A. Thickness, cross-sectional areas and RhoC. Nature 2000;406:532^ 5. ity in high-dimensional classification tasks. Stat Appl depth of invasion in the prognosis of cutaneous mela- 16. Bittner M, Meltzer P, ChenY, et al. Molecular classi- Genet Mol Biol 2004;3:article 37. noma. Ann Surg 1970;172:902 ^ 8. fication of cutaneous malignant melanoma by gene 28. von Heydebreck A, HuberW, Poustka A,Vingron M. 3. Chapman PB, Einhorn LH, Meyers M, et al. Phase III expression profiling. Nature 2000;406:536 ^ 40. Identifying splits with clear separation: a new class multicenter randomized trial of the Dartmouth regimen 17. Winnipenninckx V, Lazar V, Michiels S, et al. Gene discovery method for gene expression data. Bioinfor- versus dacarbazine in patients with metastatic mela- expression profiling of primary cutaneous melanoma matics 2001;17:S107 ^ 14. noma. J Clin Oncol 1999;17:2745 ^ 51. and clinical outcome. J Natl Cancer Inst 2006;98: 29. Gruss C, Herlyn M. Role of cadherins and matrixins 4. Middleton MR, Grob JJ, Aaronson N, et al. Random- 472 ^ 82. in melanoma. Curr Opin Oncol 2001;13:117^ 23. ized phase III study of temozolomide versus dacarba- 18. Haqq C, Nosrati M, Sudilovsky D, et al.The gene ex- 30.ThiesA,MollI,BergerJ,etal.CEACAM1expression zine in the treatment of patients with advanced pression signatures of melanoma progression. Proc in cutaneous malignant melanoma predicts the devel- metastatic malignant melanoma. J Clin Oncol 2000; Natl Acad Sci U S A 2005;102:6092 ^ 7. opment of metastatic disease. J Clin Oncol 2002;20: 18:15 8 ^ 66. 19. Bonner RF, Emmert-Buck M, Cole K, et al. Laser 2530 ^ 6. 5. Thompson J, Menzies S, Shaw H, Scolyer R, Kefford capture microdissection: molecular analysis of tissue. 31. Demierre MF, Chung C, Miller DR, Geller AC. Early R. Cutaneous melanoma. Lancet 2005;365:2004 ^ 5. Science1997;278:1481^3. detection of thick melanomas in the United States: 6. Liotta L, Petricoin E. Molecular profiling of human 20. Westphal JR, van’t Hullenaar RG, van der Laak JA, beware of the nodular subtype. Arch Dermatol 2005; cancer. Nat Rev Genet 2000;1:48 ^ 56. et al.Vascular density in melanoma xenografts corre- 141:74 5 ^ 5 0. 7. Ramaswamy S, GolubTR. DNA microarrays in clinical lates with vascular permeability factor expression but 32. Bu«ttner PG, Leiter U, Eigentler TK, Garbe C. De- oncology. J Clin Oncol 2002;20:1932^ 41. not with metastatic potential. Br J Cancer 1997;76: velopment of prognostic factors and survival in cu- 8. Liang P, Pardee AB. Analysing differential gene ex- 561 ^ 70. taneous melanoma over 25 years: an analysis of the pression in cancer. Nat Rev Cancer 2003;3:869 ^ 76. 21. Huber W, von Heydebreck A, Su«ltmann H, Central Malignant Melanoma Registry of the Ger- 9. Perou CM, Sorlie T, Eisen MB, et al. Molecular por- Poustka A, Vingron M. Variance stabilization applied manDermatologicalSociety.Cancer2005;103: traits of human breast tumours. Nature 2000;406: to microarray data calibration and to the quantifica- 616 ^ 24. 747 ^ 52. tion of differential expression. Bioinformatics 2002; 33. Bu«ttner P,Garbe C, BertzJ, et al. Primary cutaneous 10. Bhattacharjee A, Richards WG, Staunton J, et al. 18 :S96 ^ 10 4. melanoma. Optimized cutoff points of tumor thickness Classification of human lung carcinomas by mRNA ex- 22. Irizarry R, Hobbs B, Collin F, et al. Exploration, nor- and importance of Clark’s level for prognostic classifi- pression profiling reveals distinct adenocarcinoma malization, and summaries of high density oligonucle- cation. Cancer 1995;75:2499 ^ 506. subclasses.ProcNatlAcadSciUSA2001;98: otide array probe level data. Biostatistics 2003;4: 34. Meyer D, Leisch F, Hornik K. Benchmarking sup- 1379 0 ^ 5 . 249^ 64. port vector machines. Neurocomputing 2003;55: 11. Crnogorac-Jurcevic T, Efthimiou E, Nielsen T, et al. 23. Tusher V, Tibshirani R, Chu G. Significance analy- 169 ^ 8 6. Expression profiling of microdissected pancreatic sis of microarrays applied to the ionizing radiation 35. Petersen BO, Lukas J, Sorensen CS, Bartek J, Helin adenocarcinomas. Oncogene 2002;21:4587 ^ 94. response. Proc Natl Acad Sci U S A 2001;98: K. Phosphorylation of mammalian CDC6 by cyclin A/ 12. ZouTT, Selaru FM, XuY, et al. Application of cDNA 5116 ^ 21. CDK2 regulates its subcellular localization. EMBO J microarrays to generate a molecular taxonomy capa- 24. Storey JD, Tibshirani R. Statistical significance for 1999;18:396^410. ble of distinguishing between colon cancer and nor- genomewide studies. Proc Natl Acad Sci U S A 36. Fujita M,Yamada C, Goto H, et al. Cell cycle regula- mal colon. Oncogene 2002;21:4855 ^ 62. 2003;100:9440 ^5. tion of human CDC6 protein. Intracellular localization, 13. van’t Veer LJ, Dai H, van de Vijver MJ, et al. Gene 25. Cortes C, Vapink V. Support vector networks. Ma- interaction with the human mcm complex, and CDC2 expression profiling predicts clinical outcome of breast chine Learning 1995;20:273^ 97. kinase-mediated hyperphosphorylation. J Biol Chem cancer. Nature 2002;415:530 ^ 6. 26. BrownMPS,GroundyWN,LinD,etal.Knowledge 1999;274:25927 ^ 32. 14. Ramaswamy S, Ross KN, Lander ES, Golub TR. A based analysis of microarray gene expression data by 37. Hansel DE, Dhara S, Huang RC, et al. CDC2/CDK1 molecular signature of metastasis in primary solid using support vector machines. Proc Natl Acad Sci expression in esophageal adenocarcinoma and tumors. Nat Genet 2003;33:49 ^ 54. U S A 2000;97:262 ^ 7. precursor lesions serves as a diagnostic and cancer

Clin Cancer Res 2007;13(3) February1,2007 814 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Molecular Signatures of Malignant Melanoma

progression marker and potential novel drug target. mammalian cells. Proc Natl Acad Sci U S A 1998;95: 48. Haass NK, Smalley KSM, Li L, Herlyn M. Ad- Am J Surg Pathol 2005;29:390 ^ 9. 3603^8. hesion, migration and communication in melano- 38. Hanahan D, Weinberg RA. The hallmarks of cancer. 44. Murphy N, Ring M, Heffron CC, et al. p16INK4A, cytes and melanoma. Pigment Cell Res 2005;18: Cell 2000;100:57 ^ 70. CDC6, and MCM5: predictive biomarkers in cervical 15 0 ^ 9. 39. Weber BL. Cancer genomics. Cancer Cell 2002;1: preinvasive neoplasia and cervical cancer. J Clin 49. Hsu MY, Shih DT, Meier FE, et al. Adenoviral 37^47. Pathol 2005;58:525^ 34. gene transfer of beta3integrin subunit induces 40. Chin L. Modeling malignant melanoma in mice: 45. Senderowicz AM. Targeting cell cycle and apoptosis conversion from radial to vertical growth phase in pathogenesis and maintenance. Oncogene 1999;18: for the treatment of human malignancies. Curr Opin primary human melanoma. Am J Pathol 1998;153: 5304^10. Cell Biol 2004;16:670 ^ 8. 1435 ^ 42. 41. Chin L. The genetics of malignant melanoma: les- 46. Laoukili J, Kooistra MR, Bras A, et al. FoxM1is re- 50. Trisciuoglio D, Desideri M, Ciuffreda L, et al. Bcl-2 sons from mouse and man. Nat Rev Cancer 2003;3: quired for execution of the mitotic programme and overexpression in melanoma cells increases tumor 559 ^ 70. chromosome stability. Nat Cell Biol 2005;7:126^ 36. progression-associated properties and in vivo tumor 42. Sharpless E, Chin L.The INK4a/ARF locus and mel- 47. Clark GM, Allred DC, Hilsenbeck SG, et al. Mitosin growth. J Cell Physiol 2005;205:414^ 21. anoma. Oncogene 2003;22:3092 ^ 8. (a new proliferation marker) correlates with clinical 51. Lomuto M, Calabrese P, Giuliani A. Prognostic signs 43. Yan Z, DeGregori J, Shohet R, et al. Cdc6 is regu- outcome in node-negative breast cancer. Cancer Res in melanoma: state of the art. J Eur Acad Dermatol lated by E2F and is essential for DNA replication in 1997;57:5505 ^ 8. Venereol 2004;18:291^300.

www.aacrjournals.org 815 Clin Cancer Res 2007;13(3) February 1,2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Gene Expression Signatures for Tumor Progression, Tumor Subtype, and Tumor Thickness in Laser-Microdissected Melanoma Tissues

Jochen Jaeger, Dirk Koczan, Hans-Juergen Thiesen, et al.

Clin Cancer Res 2007;13:806-815.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/3/806

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2007/02/01/13.3.806.DC1

Cited articles This article cites 49 articles, 15 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/3/806.full#ref-list-1

Citing articles This article has been cited by 19 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/3/806.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/13/3/806. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.