Access the full text.
Sign up today, get DeepDyve free for 14 days.
G. Long, A. Menzies, A. Nagrial, L. Haydu, Anne Hamilton, G. Mann, T. Hughes, J. Thompson, R. Scolyer, R. Kefford (2011)
Prognostic and clinicopathologic associations of oncogenic BRAF in metastatic melanoma.Journal of clinical oncology : official journal of the American Society of Clinical Oncology, 29 10
K. Montone, P. Belle, R. Elenitsas, David Elder (1997)
Proto-oncogene c-kit expression in malignant melanoma: protein loss with tumor progression.Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc, 10 9
Wen Cheng, Xiufang Ren, Chuanbao Zhang, Jinquan Cai, Yang Liu, Sheng Han, A. Wu (2016)
Bioinformatic profiling identifies an immune-related risk signature for glioblastomaNeurology, 86
K. Pilipow, A. Roberto, M. Roederer, T. Waldmann, D. Mavilio, E. Lugli (2015)
IL15 and T-cell Stemness in T-cell-Based Cancer Immunotherapy.Cancer research, 75 24
Glynn Dennis, Brad Sherman, Douglas Hosack, Jun Yang, Wei Gao, H. Lane, R. Lempicki (2003)
DAVID: Database for Annotation, Visualization, and Integrated DiscoveryGenome Biology, 4
Willy Hugo, J. Zaretsky, Lu Sun, Chunying Song, B. Moreno, S. Hu-Lieskovan, B. Berent-Maoz, Jia Pang, B. Chmielowski, Grace Cherry, E. Seja, Shirley Lomeli, X. Kong, M. Kelley, J. Sosman, Douglas Johnson, A. Ribas, R. Lo (2017)
Genomic and Transcriptomic Features of Response to Anti-PD-1 Therapy in Metastatic MelanomaCell, 168
Nikenza Viceconte, Marie-Sophie Dheur, E. Majerová, C. Pierreux, J. Baurain, N. Baren, A. Decottignies (2017)
Highly Aggressive Metastatic Melanoma Cells Unable to Maintain Telomere Length.Cell reports, 19 12
(2007)
Academic Editor: Jonathan Rees, University of Edinburgh, United Kingdom
Craig April, Brandy Klotzle, Thomas Royce, E. Wickham-Garcia, T. Boyaniwsky, John Izzo, D. Cox, W. Jones, Renee Rubio, K. Holton, U. Matulonis, John Quackenbush, Jian-Bing Fan (2009)
Whole-Genome Gene Expression Profiling of Formalin-Fixed, Paraffin-Embedded Tissue SamplesPLoS ONE, 4
S. Ratuapli, A. Bharucha, J. Noelting, Doris Harvey, A. Zinsmeister (2013)
Phenotypic identification and classification of functional defecatory disorders using high-resolution anorectal manometry.Gastroenterology, 144 2
B. Weir, M. Woo, G. Getz, S. Perner, L. Ding, R. Beroukhim, William Lin, M. Province, A. Kraja, L. Johnson, K. Shah, Mitsuo Sato, Roman Thomas, J. Barletta, I. Borecki, S. Broderick, A. Chang, Derek Chiang, L. Chirieac, Jeonghee Cho, Y. Fujii, A. Gazdar, T. Giordano, H. Greulich, M. Hanna, B. Johnson, M. Kris, A. Lash, Ling Lin, N. Lindeman, E. Mardis, J. McPherson, J. Minna, M. Morgan, M. Nadel, M. Orringer, John Osborne, B. Ozenberger, A. Ramos, James Robinson, J. Roth, V. Rusch, H. Sasaki, F. Shepherd, C. Sougnez, M. Spitz, M. Tsao, David Twomey, R. Verhaak, G. Weinstock, D. Wheeler, W. Winckler, A. Yoshizawa, Soyoung Yu, M. Zakowski, Qunyuan Zhang, D. Beer, I. Wistuba, M. Watson, L. Garraway, M. Ladanyi, W. Travis, W. Pao, M. Rubin, S. Gabriel, R. Gibbs, H. Varmus, R. Wilson, E. Lander, M. Meyerson (2007)
Characterizing the cancer genome in lung adenocarcinomaNature, 450
F. Donskov (2013)
Immunomonitoring and prognostic relevance of neutrophils in clinical trials.Seminars in cancer biology, 23 3
K. Harbst, J. Staaf, M. Lauss, Anna Karlsson, A. Måsbäck, I. Johansson, P. Bendahl, J. Vallon-Christersson, Therese Törngren, Henrik Ekedahl, J. Geisler, M. Höglund, M. Ringnér, L. Lundgren, K. Jirström, H. Olsson, C. Ingvar, Å. Borg, H. Tsao, G. Jönsson (2012)
Molecular Profiling Reveals Low- and High-Grade Forms of Primary MelanomaClinical Cancer Research, 18
R. Mortarini, A. Scarito, D. Nonaka, M. Zanon, I. Bersani, Elisabetta Montaldi, E. Pennacchioli, R. Patuzzo, M. Santinami, A. Anichini (2005)
Constitutive expression and costimulatory function of LIGHT/TNFSF14 on human melanoma cells and melanoma-derived microvesicles.Cancer research, 65 8
R. Zhang, J. Price, G. Schackert, K. Itoh, I. Fidler (1991)
Malignant potential of cells isolated from lymph node or brain metastases of melanoma patients and implications for prognosis.Cancer research, 51 8
(1997)
Prognostic Model for Primary Melanoma
E. Gaude, C. Frezza (2016)
Tissue-specific and convergent metabolic transformation of cancer correlates with metastatic potential and patient survivalNature Communications, 7
Debra Black, Laura Brockway-Lunardi (2013)
The Melanoma Research Alliance: The Power of Patient Advocacy to Accelerate Research and Novel TherapiesCancer Immunology Research, 1
M. Harris, Jennifer Clark, A. Ireland, J. Lomax, M. Ashburner, M. Ashburner, R. Foulger, R. Foulger, K. Eilbeck, K. Eilbeck, S. Lewis, S. Lewis, B. Marshall, B. Marshall, C. Mungall, C. Mungall, J. Richter, J. Richter, G. Rubin, Gerald Rubin, J. Blake, C. Bult, M. Dolan, H. Drabkin, J. Eppig, Hill Dp, L. Ni, M. Ringwald, R. Balakrishnan, R. Balakrishnan, J. Cherry, J. Cherry, K. Christie, K. Christie, M. Costanzo, M. Costanzo, S. Dwight, S. Dwight, S. Engel, S. Engel, D. Fisk, D. Fisk, J. Hirschman, J. Hirschman, Eurie Hong, Eurie Hong, R. Nash, R. Nash, A. Sethuraman, A. Sethuraman, C. Theesfeld, C. Theesfeld, D. Botstein, D. Botstein, K. Dolinski, K. Dolinski, B. Feierbach, B. Feierbach, T. Berardini, T. Berardini, S. Mundodi, S. Mundodi, S. Rhee, S. Rhee, R. Apweiler, D. Barrell, E. Camon, E. Dimmer, Lee, R. Chisholm, P. Gaudet, P. Gaudet, W. Kibbe, W. Kibbe, R. Kishore, R. Kishore, E. Schwarz, E. Schwarz, P. Sternberg, P. Sternberg, M. Gwinn, L. Hannick, J. Wortman, M. Berriman, Wood, N. Delacruz, N. Delacruz, P. Tonellato, P. Tonellato, P. Jaiswal, P. Jaiswal, T. Seigfried, T. Seigfried, R. White, R. White (2004)
Gene Ontology Consortium: The Gene Ontology (GO) database and informatics resourceNucleic acids research, 32 Database issue
C. Barker, M. Postow, Shaheer Khan, K. Beal, P. Parhar, Y. Yamada, N. Lee, J. Wolchok (2013)
Concurrent Radiotherapy and Ipilimumab Immunotherapy for Patients with MelanomaCancer Immunology Research, 1
A. Tarhini, Yan Lin, Hui-Min Lin, Priyanka Vallabhaneni, C. Sander, W. LaFramboise, L. Hamieh (2017)
Expression profiles of immune-related genes are associated with neoadjuvant ipilimumab clinical benefitOncoImmunology, 6
K. Ransohoff, Wenting Wu, H. Cho, H. Chahal, Yuan Lin, H. Dai, C. Amos, Jeffrey Lee, Jean Tang, D. Hinds, Jiali Han, Q. Wei, K. Sarin (2017)
Two-stage genome-wide association study identifies a novel susceptibility locus associated with melanomaOncotarget, 8
(2008)
Improving melanoma classification by integrating genetic and morphologic featuresPLoS Medicine, 5
Alexander Huth, Shinji Nishimoto, A. Vu, J. Gallant (2012)
A Continuous Semantic Space Describes the Representation of Thousands of Object and Action Categories across the Human BrainNeuron, 76
Aniruddha Chatterjee, P. Stockwell, Euan Rodger, Matthew Parry, M. Eccles (2016)
scan_tcga tools for integrated epigenomic and transcriptomic analysis of tumor subgroups.Epigenomics, 8 10
Ash Alizadeh, M. Eisen, R. Davis, Chishieva Ma, I. Lossos, A. Rosenwald, Jennifer Boldrick, H. Sabet, T. Tran, Xin Yu, J. Powell, Liming Yang, G. Marti, T. Moore, J. Hudson, Li-Sheng Lu, D. Lewis, R. Tibshirani, G. Sherlock, W. Chan, T. Greiner, D. Weisenburger, J. Armitage, R. Warnke, R. Levy, W. Wilson, M. Grever, J. Byrd, D. Botstein, P. Brown, L. Staudt (2000)
Distinct types of diffuse large B-cell lymphoma identified by gene expression profilingNature, 403
C. Lovly, K. Dahlman, Laurel Fohn, Zengliu Su, D. Dias-Santagata, Donna Hicks, Donald Hucks, Elizabeth Berry, Charles Terry, Markeesa Duke, Yingjun Su, T. Sobolik-Delmaire, A. Richmond, Mark Kelley, Cindy Vnencak-Jones, A. Iafrate, J. Sosman, W. Pao (2012)
Routine Multiplex Mutational Profiling of Melanomas Enables Enrollment in Genotype-Driven Therapeutic TrialsPLoS ONE, 7
A. Subramanian, P. Tamayo, V. Mootha, Sayan Mukherjee, B. Ebert, Michael Gillette, A. Paulovich, S. Pomeroy, T. Golub, E. Lander, J. Mesirov (2005)
Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profilesProceedings of the National Academy of Sciences of the United States of America, 102
Todd Golub, Todd Golub, D. Slonim, Pablo Tamayo, Christine Huard, Michelle Gaasenbeek, J. Mesirov, Hilary Coller, M. Loh, James Downing, Michael Caligiuri, C. Bloomfield, Eric Lander (1999)
Molecular classification of cancer: class discovery and class prediction by gene expression monitoring.Science, 286 5439
S. Champiat, L. Dercle, S. Ammari, C. Massard, A. Hollebecque, S. Postel-Vinay, N. Chaput, A. Eggermont, A. Marabelle, J. Soria, C. Ferté (2016)
Hyperprogressive Disease Is a New Pattern of Progression in Cancer Patients Treated by Anti-PD-1/PD-L1Clinical Cancer Research, 23
L. Kwong, J. Costello, Huiyun Liu, Shan Jiang, Timothy Helms, Aliete Langsdorf, David Jakubosky, G. Genovese, F. Muller, Joseph Jeong, Ryan Bender, Gerald Chu, K. Flaherty, J. Wargo, J. Collins, L. Chin (2012)
Oncogenic NRAS signaling differentially regulates survival and proliferation in melanomaNature Medicine, 18
Jun Yao, O. Caballero, Ying Huang, Calvin Lin, D. Rimoldi, A. Behren, J. Cebon, M. Hung, J. Weinstein, R. Strausberg, Qi Zhao (2016)
Altered Expression and Splicing of ESRP1 in Malignant Melanoma Correlates with Epithelial–Mesenchymal Status and Tumor-Associated Immune Cytolytic ActivityCancer Immunology Research, 4
Li Wang, Wei Zhou, Y. Zhong, Yongbao Huo, P. Fan, Sudong Zhan, Jun Xiao, Xin Jin, S. Gou, T. Yin, Heshui Wu, Tao Liu (2017)
Overexpression of G protein-coupled receptor GPR87 promotes pancreatic cancer aggressiveness and activates NF-κB signaling pathwayMolecular Cancer, 16
I. Lossos, D. Czerwinski, Ash Alizadeh, M. Wechser, R. Tibshirani, D. Botstein, R. Levy (2004)
Prediction of survival in diffuse large-B-cell lymphoma based on the expression of six genes.The New England journal of medicine, 350 18
D. Sia, Y. Jiao, Iris Martinez-Quetglas, O. Kuchuk, C. Villacorta-Martin, M. Moura, J. Putra, Genís Campreciós, L. Bassaganyas, N. Akers, B. Losic, S. Waxman, S. Thung, V. Mazzaferro, M. Esteller, S. Friedman, M. Schwartz, A. Villanueva, J. Llovet (2017)
Identification of an Immune-specific Class of Hepatocellular Carcinoma, Based on Molecular Features.Gastroenterology, 153 3
Zichen Wang, Caroline Monteiro, Kathleen Jagodnik, Nicolas Fernandez, Gregory Gundersen, A. Rouillard, S. Jenkins, Axel Feldmann, Kevin Hu, Michael McDermott, Qiaonan Duan, N. Clark, Matthew Jones, Y. Kou, T. Goff, Holly Woodland, Fabio Amaral, G. Szeto, Oliver Fuchs, S. Rose, Shvetank Sharma, U. Schwartz, Xabier Bausela, Maciej Szymkiewicz, Vasileios Maroulis, A. Salykin, C. Barra, Candice Kruth, N. Bongio, Vaibhav Mathur, Radmila Todoric, Udi Rubin, A. Malatras, Carl Fulp, J. Galindo, R. Motiejūnaitė, Christoph Jüschke, Philip Dishuck, K. Lahl, Mohieddin Jafari, S. Aibar, A. Zaravinos, Linda Steenhuizen, Lindsey Allison, Pablo Gamallo, Fernando Segura, Tyler Devlin, V. Perez-Garcia, Avi Ma’ayan (2016)
Extraction and analysis of signatures from the Gene Expression Omnibus by the crowdNature Communications, 7
E. Allen, Diana Miao, B. Schilling, S. Shukla, C. Blank, L. Zimmer, A. Sucker, U. Hillen, M. Foppen, S. Goldinger, J. Utikal, J. Hassel, B. Weide, K. Kaehler, C. Loquai, P. Mohr, R. Gutzmer, R. Dummer, S. Gabriel, Catherine Wu, D. Schadendorf, L. Garraway (2015)
Genomic correlates of response to CTLA-4 blockade in metastatic melanomaScience, 350
G. Jönsson, C. Busch, S. Knappskog, J. Geisler, H. Miletic, M. Ringnér, J. Lillehaug, Å. Borg, P. Lønning (2010)
Gene Expression Profiling–Based Identification of Molecular Subtypes in Stage IV Melanomas with Different Clinical OutcomeClinical Cancer Research, 16
S. Broekaert, R. Roy, I. Okamoto, J. Oord, J. Bauer, C. Garbe, R. Barnhill, K. Busam, A. Cochran, M. Cook, D. Elder, S. McCarthy, M. Mihm, D. Schadendorf, R. Scolyer, A. Spatz, B. Bastian (2010)
Genetic and morphologic features for melanoma classificationPigment Cell & Melanoma Research, 23
BioBusiness Briefs (2010)
Trial watch: Ipilimumab success in melanoma provides boost for cancer immunotherapyNature Reviews Drug Discovery, 9
K. Yoshihara, Maria Shahmoradgoli, Emmanuel Martinez, Rahulsimham Vegesna, Hoon Kim, Wandaliz Torres-García, V. Treviño, Hui Shen, P. Laird, D. Levine, S. Carter, G. Getz, K. Stemke‐Hale, G. Mills, R. Verhaak (2013)
Inferring tumour purity and stromal and immune cell admixture from expression dataNature Communications, 4
J. Luke, K. Flaherty, A. Ribas, G. Long (2017)
Targeted agents and immunotherapies: optimizing outcomes in melanomaNature Reviews Clinical Oncology, 14
S. Delić, S. Delić, N. Lottmann, K. Jetschke, G. Reifenberger, M. Riemenschneider, M. Riemenschneider (2012)
Identification and functional validation of CDH11, PCSK6 and SH3GL3 as novel glioma invasion‐associated candidate genesNeuropathology and Applied Neurobiology, 38
Hindawi Journal of Oncology Volume 2020, Article ID 7526204, 13 pages https://doi.org/10.1155/2020/7526204 Research Article Identification of a Gene-Related Risk Signature in Melanoma Patients Using Bioinformatic Profiling 1 2,3 4,5,6 4,5 1 Jing Wang , Peng-Fei Kong, Hai-Yun Wang, Di Song, Wen-Qing Wu, 1 1 1 1 1 1 Hang-Cheng Zhou, Hai-Yan Weng, Ming Li, Xin Kong, Bo Meng, Zong-Ke Chen, 1 1 4,5 Jing-Jing Chen, Chuan-Ying Li , and Jian-Yong Shao Department of Pathology, e First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, Anhui 230001, China Department of Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, China Department of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and erapy, Sun Yat-sen University Cancer Center, No. 651 Dongfeng East Road, Guangzhou 510060, China Department of Molecular Diagnostics, Sun Yat-sen University Cancer Center, Guangzhou 510060, China Department of Heart Medicine, Guangzhou Institute of Pediatrics, Guangzhou Women and Children’s Medical Centre, Guangzhou Medical University, Guangzhou, Guangdong, China Correspondence should be addressed to Chuan-Ying Li; lcy12224@rjh.com.cn and Jian-Yong Shao; shaojy@sysucc.org.cn Received 19 July 2019; Accepted 21 January 2020; Published 29 April 2020 Academic Editor: Akira Hara Copyright © 2020 Jing Wang et al. +is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Introduction. Gene signature has been used to predict prognosis in melanoma patients. Meanwhile, the efficacy of immunotherapy was correlated with particular genes expression or mutation. In this study, we systematically explored the gene expression pattern in the melanoma-immune microenvironment and its relationship with prognosis. Methods. A cohort of 122 melanoma cases with whole-genome microarray expression data were enrolled from the Gene Expression Omnibus (GEO) database. +e findings were validated using +e Cancer Genome Atlas (TCGA) database. A principal component analysis (PCA), gene set enrichment analysis (GSEA), and gene oncology (GO) analysis were performed to explore the bioinformatic implications. Results. Different gene expression patterns were identified according to the clinical stage. All eligible gene sets were analyzed, and the 8 genes (GPR87, KIT, SH3GL3, PVRL1, ATP1B1, CDAN1, FAU, and TNFSF14) with the greatest prognostic impact on melanoma. A gene-related risk signature was developed to distinguish patients with a high or low risk of an unfavorable outcome, and this signature was validated using the TCGA database. Furthermore, the prognostic significance of the signature between the classified subgroups was verified as an independent prognostic predictor of melanoma. Additionally, the low-risk melanoma patients presented an enhanced immune phenotype compared to that of the high-risk gene signature patients. Conclusions. +e gene pattern differences in melanoma were profiled, and a gene signature that could independently predict melanoma patients with a high risk of poor survival was established, highlighting the relationship between prognosis and the local immune response. mRNA expression profiling studies, melanoma can be di- 1. Introduction vided into molecular subtypes, and several subtypes share To date, many advancements in melanoma have elucidated clinical properties and gene expression patterns [3, 4]. Since the positive and negative relationships between various the survival rate of melanoma patients does not significantly clinicopathological features and prognosis. For instance, improve after standard treatment, the novel approach of metastasis accounts for over 90% of cancer-specific mortality immunotherapy is currently under intensive investigation in melanoma [1, 2]. According to recent whole-genome [5, 6]. In addition, several gene patterns in melanoma have 2 Journal of Oncology been reported to predict the strength of the antitumor re- database (Database for Annotation, Visualization and Inte- sponse [7, 8], further highlighting the importance of precise gration Discovery, http://david.abcc.ncifcrf.gov/) was used to conduct a functional enrichment analysis in our study [14, 15]. gene signature stratification in predicting immunotherapy outcomes. However, only a few studies have systematically Furthermore, the normalized enrichment score (NES) and false explored the gene expression pattern in the melanoma- discovery rate (FDR) were applied to determine the significant immune microenvironment and its relationship with differences. A p-value <0.05 was set as the threshold. prognosis. Altogether, a better understanding of the mo- lecular characteristics of melanoma is highly significant. 2.4. Statistical Analysis. Using the RNA-Seq database, the In this study, we profiled the gene expression patterns in log 2 expression values were calculated for each probe [16]. 122 melanoma patients using whole-genome expression data For genes with several probes, the median was calculated for from the Gene Expression Omnibus (GEO) database. Dis- further analysis. A univariate Cox regression analysis was tinct degrees of phenotype enrichment were established performed to evaluate the significance of the prognostic based on the clinical stage. Using the enriched gene sig- value of the genes in melanoma. We found 8 genes that were nature in melanoma, we found a gene-related risk signature highly correlated with the OS (p< 0.01), and these genes by profiling the whole gene set, and this signature was were either associated with risk or protective based on their subsequently validated using +e Cancer Genome Atlas hazard ratio (HR). An 8-gene risk signature model was (TCGA) database. Our gene-related risk signature can in- established for the prediction of survival, a univariate Cox dependently identify melanoma patients at high risk of regression analysis was conducted and a linear combination unfavorable clinical outcomes, and the expression intensity of their expression levels weighted using the regression of immune-related genes is severely reduced in these pa- coefficients was determined with the OS as the dependent tients, thereby indicating that survival is closely associated variable [17]. Next, the melanoma patients from both the with the melanoma-immune microenvironment. GEO and TCGA datasets were divided into high- and low- risk groups based on their median protection value. Both 2. Materials and Methods univariate and multivariate Cox regression analyses were performed to identify the independent prognostic factors. 2.1. Patient Samples. In total, 581 melanoma samples from the Gene Expression Omnibus (GEO) and +e Cancer +e primary endpoint was calculated using the Kaplan–Meier method, and the survival curves were com- Genome Atlas (TCGA) database were included in our study (Supplementary Tables 1 and 2) [9, 10]. +e GEO and TCGA pared using a 2-tailed log-rank test. Additionally, the dif- gene expression profiles (RNA-Seq expression) and corre- ferences in the clinicopathological features between the groups were evaluated using Fisher’s exact test or χ sponding clinical metadata were accessed from the GEO tests. All (https://www.ncbi.nlm.nih.gov/geo/) and TCGA (https:// statistical analyses were conducted using SPSS software tcga-data.nci.nih.gov/tcga/dataAccess-Matrix.htm) public (Version 19.0, SPSS Inc.) and GraphPad Prism (Version 5.0, access databases released before May 20, 2017. +e overall GraphPad Software Inc.). A 2-sided p-value < 0.05 was considered statistically significant. +e PCA and the gen- survival (OS) was defined from the date of diagnosis until death or the end of follow-up. eration of the heatmap and Circos diagrams were performed using R software (Version 3.4.2). 2.2. Standard Protocol Approval, Registration, and Patient 3. Results Consent. +is study was approved by the Ethics Committee and Institutional Review Board of SYSUCC. All enrolled 3.1. Enhanced Gene Expression in Primary and Metastatic patients signed informed consent forms. Melanoma. We analyzed 122 primary and metastatic mel- anoma cases obtained from the GEO database (GSE59455) 2.3. Principal Components Analysis (PCA), Gene Set En- using mRNA expression and clinical data (Supplementary richment Analysis (GSEA), and Gene Oncology (GO) Analysis. Table 1). A clinical diagnosis was achieved and defined in 37 For each subject we used principal components analysis (PCA) and 85 patients in primary and metastatic status, respec- to recover a low-dimensional semantic space from category tively. We downloaded all gene sets (hallmark and C1 to C7) model weights and classify the gene signature patterns in the from the Molecular Signatures Database (http://software. patients. As previously studies reported, participants’ scores in broadinstitute.org/gsea/downloads.jsp) and combined the all assessments were entered into the PCA with maximum gene sets to obtain a total gene set containing more than fluctuations [11]. Next, we use the degrees of freedom signif- 20000 genes [18]. icance threshold (p< 0.05 for multiple comparisons uncor- Because the clinical and biological differences have been rected) to select all voxels that the model predicts significantly well established (Supplementary Table 3), we objected to [12]. We then applied PCA to the category model weights of the further explore the different gene patterns between primary selected voxels. A GSEA (http://www.broadinstitute.org/gsea/ and metastatic melanoma. All gene sets were used to per- index.jsp) was conducted to determine whether the identified form a GSEA analysis. A significantly different enrichment sets of genes significantly differed between the groups [13]. A in the DNA binding-related, metastasis-related, and other GO enrichment analysis of the differentially expressed genes in gene sets was observed (Figures 1(a), 1(b), and Supple- the gene expression network was conducted. +e DAVID mentary Tables 4 and 5), revealing an entirely different gene Journal of Oncology 3 RNA-dependent- Telomere maintenance Telomeric DNA binding DNA biosynthetic process NES = 2.2989 NES = 2.3817 NES = 2.2243 FDR = 0.0021 FDR = 0.0010 FDR = 0.0034 Metastatic Metastatic Metastatic Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Enrichment profile Hits Ranking metric scores (a) Jaeger metastasis Immortalized by HPV31 Huper brast basal NES = –2.2788 NES = –2.2767 NES = –2.2564 FDR = 0.0292 FDR = 0.0146 FDR = 0.0139 Metastatic Metastatic Metastatic Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Enrichment profile Hits Ranking metric scores (b) Primary Primary Metastatic Metastatic (c) (d) Figure 1: Different gene expression patterns between primary and metastatic melanoma. (a, b) Gene set enrichment analysis (GSEA) was performed to compare the gene expression between metastatic and primary tumors. FDR � false discovery rate; NES � normalized en- richment score. (c) Principal components analysis of the whole genome between primary and metastatic melanoma. (d) Principal components analysis of enriched genes between primary and metastatic melanoma. Ranked list metric Enrichment Ranked list metric Enrichment (Signal2Noise) score (ES) (Signal2Noise) score (ES) 4 Journal of Oncology TNFSF14 CDAN1 FAU GPR87 ATP1B1 p < 0.0001 GALNT8 EREG EIF2AK4 DNAJB6 KLF5 GPX1 KIT 0 5 10 15 20 25 SH3GL3 Time (years) PVRL1 High risk (n = 61) Low risk (n = 61) –0.3 –0.2 –0.1 0 0.1 0.2 0.3 Regression coefficient (a) (b) p < 0.0001 0 5 10 15 20 25 30 35 Time (years) High risk (n = 230) Low risk (n = 229) (c) Figure 2: An 8-gene local gene signature in patients with melanoma. (a) +e dashed lines represent an absolute regression coefficient of ±0.05. +e prediction model is based on the weighted expression of eight genes and is expressed by the following equation: Protection score � (0.202 × TNFSF14) + (0.091 × CDAN1) + (0.081 × FAU) + (0.071 × GPR87) + (0.052 × ATP1B1) + (−0.094 × KIT) + (−0.179 × SH3GL3) + (−0.250 × PVRL1). (b) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in the GEO database. (c) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in the TCGA database. signature between the two groups. As shown in Figure 1(c), (Supplementary Table 6, p< 0.01). +en, the risk score method was used to establish a risk signature for melanoma the PCA based on the whole-genome expression data showed a different intertwined pattern. Furthermore, a PCA patients based on the gene expression levels [17]. As pre- based on the enrichment gene set data showed a relatively sented in Figure 2(a), we ranked the genes based on their different distribution pattern (Figure 1(d)). Primary mela- predictive power (regression coefficients). We excluded noma was distributed on the left side, while metastatic several genes with relatively low predictive power melanoma was distributed on the right side, indicating (−0.05< regression coefficient< 0.05). Finally, eight genes remarkably distinct gene expression patterns between the (GPR87, KIT, SH3GL3, PVRL1, ATP1B1, CDAN1, FAU, clinical stages. and TNFSF14) were identified to be closely associated with the OS in melanoma. In addition, all identified genes were of the following 2 types: risky or protective. An HR> 1 was 3.2. Identification of a Local Gene Signature to Predict Prog- defined as risky (GPR87, KIT, SH3GL3, and PVRL1), and an nosis in Melanoma Patients. Considering the enrichment HR< 1 was defined as protective (ATP1B1, CDAN1, FAU, gene expression in primary and metastatic melanoma, we and TNFSF14). attempted to establish a local gene signature as a predictor of +e prediction model is based on the weighted ex- prognosis. Subsequently, we performed a univariate Cox pression of eight genes and is expressed by the following regression analysis to explore the prognostic value of these equation: protection value score � (0.202 × TNFSF14) + enriched genes. In our study, fourteen genes (EREG, (0.091 × CDAN1) + (0.081 × FAU) + (0.071 × GPR87) + GALNT8, GPR87, KIT, KLF5, SH3GL3, PVRL1, ATP1B1, (0.052 × ATP1B1) + (−0.094 × KIT) + (−0.179 × SH3GL3) + CDAN1, DNAJB6, EIF2AK4, FAU, GPX1, and TNFSF14) (−0.250 × PVRL1). All cases were divided into high-risk were observed to predict survival in melanoma (n � 61) and low-risk (n � 61) subgroups based on the Overall survival (%) Overall survival (%) Journal of Oncology 5 Table 1: Multivariate Cox regression analysis of clinicopathological factors and overall survival using the TCGA database. Univariate Cox regression Multivariate Cox regression Variable HR p value HR p value Age Increasing years 1.748 0.001 1.517 0.018 Sex Female vs. male 0.784 0.979 BRAF status Mutation vs. wild-type 1.174 0.035 1.142 0.089 NRAS status Mutation vs. wild-type 0.957 0.580 Clinical stage III + IV vs. <III 1.023 0.001 1.023 0.001 Local gene signature High-risk vs. low-risk 1.318 <0.001 1.274 0.001 TCGA: +e Cancer Genome Atlas. HR: hazard ratio. 100 100 80 80 60 60 p < 0.0001 p = 0.1081 40 40 20 20 0 0 05 10 15 20 25 024 68 10 Time (years) Time (years) High risk (n = 33) High risk (n = 28) Low risk (n = 52) Low risk (n = 9) (a) (b) 100 100 80 80 60 60 p = 0.0085 p = 0.0012 40 40 20 20 05 10 15 20 25 30 35 05 10 15 20 Time (years) Time (years) High risk (n = 147) High risk (n = 81) Low risk (n = 135) Low risk (n = 95) (c) (d) Figure 3: Continued. Overall survival (%) Overall survival (%) Overall survival (%) Overall survival (%) 6 Journal of Oncology ∗∗∗ ∗∗ 2 5 –1 –5 Metastatic Primary <III III/IV (e) (f) Figure 3: Associations between the local gene signature and clinicopathological features in melanoma. (a) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in metastatic melanoma patients (GEO database). (b) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in primary melanoma patients (GEO database). (c) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in AJCC stage I/II melanoma patients (TCGA database). (d) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in stage III/IV melanoma patients (TCGA database). (e) Associations between the protection value and the clinicopathological features (Primary vs. Metastatic). (f) Associations between the protection value and the clinicopathological features (Stage I/II vs. Stage III/IV). BRAF: wild-type BRAF: mutation 100 100 80 80 60 60 p < 0.0001 p = 0.0017 40 40 20 20 0 0 05 10 15 20 25 05 10 15 20 25 Time (years) Time (years) High risk (n = 47) High risk (n = 14) Low risk (n = 44) Low risk (n = 17) (a) (b) NRAS: wild-type NRAS: mutation 100 100 60 60 p < 0.0001 p = 0.2267 40 40 0 0 05 10 15 20 25 05 10 15 20 25 Time (years) Time (years) High risk (n = 54) High risk (n = 8) Low risk (n = 41) Low risk (n = 19) (c) (d) Figure 4: Continued. Overall survival (%) Overall survival (%) Protection value Overall survival (%) Overall survival (%) Protection value Journal of Oncology 7 NRAS: wild-type NRAS: mutation p = 0.0005 p = 0.0350 40 40 20 20 0 0 010 20 30 010 20 30 Time (years) Time (years) High risk (n = 176) High risk (n = 53) Low risk (n = 156) Low risk (n = 74) (e) (f) Figure 4: Application of the local gene signature in stratified melanoma cohorts. (a) Survival curves of overall survival in high- and low-risk groups classified by the local gene signature in BRAF wild-type melanoma patients (GEO database); (b) survival curves of overall survival in high- and low-risk groups classified by the local gene signature in BRAF mutation melanoma patients (GEO database); (c) survival curves of overall survival in high- and low-risk groups classified by the local gene signature in NRAS wild-type melanoma patients (GEO database); (d) survival curves of overall survival in high- and low-risk groups classified by the local gene signature in NRAS mutation melanoma patients (GEO database); (e) survival curves of overall survival in high- and low-risk groups classified by the local gene signature in NRAS wild-type melanoma patients (TCGA database); (f) survival curves of overall survival in high- and low-risk groups classified by the local gene signature in NRAS mutation melanoma patients (TCGA database). Activation of immune Activation of immune response innate response Adaptive immune response NES = 2.0281 NES = 1.9214 NES = 1.6910 FDR < 0.0001 FDR = 0.0140 FDR = 0.0269 Low risk Low risk Low risk Rank in ordered dataset Rank in ordered dataset Rank in ordered dataset Enrichment profile Hits Ranking metric scores (a) Primary immunodeficiency Adaptive immune response syndrome based on immune receptors NES = 1.6616 NES = 1.9365 FDR = 0.0307 FDR = 0.0039 Low risk Low risk Rank in ordered dataset Rank in ordered dataset Enrichment profile Hits Ranking metric scores (b) Figure 5: Continued. Ranked list metric Enrichment (Signal2Noise) score (ES) Overall survival (%) Ranked list metric Enrichment (Signal2Noise) score (ES) Overall survival (%) CXCR4 CTLA4 HLA-DO8 SLA2 TRIM38 HLA-DOA PSMB8 MMP9 IL7R NKG7 PSMB9 TNFRSF10B SNX27 RUNX1 MSH3 ITK 8 Journal of Oncology Low risk High risk Gender Age at diagnosis Mutation status MMP9 IFITM4P IFIT3 AIRE NKTR HLA-DOA HLA-DOB CD320 HLA-DRB5 CD19 CD2 TNFRSF18 IL7R RUNX1 TNFRSF10B SLA2 CTLA4 6 SLAMF6 Gender Mutation status Female Wild-type Male BRAF mutation Age at diagnosis NRAS mutation <60 ≥60 (c) Log foldchange –2 –1 GO Terms Immune response Endosome membrane Protein binding (d) Figure 5: Immune signature differences between high- and low-risk groups in melanoma. (a, b) Significant enrichment of the immune- related phenotype in low-risk patients compared to high-risk patients. TGGA � +e Cancer Genome Atlas; FDR � false discovery rate; NES � normalized enrichment score; (c) associations between the protection value and the clinicopathological features and immune-related genes; (d) Circos diagram of the gene expression involved in immune response pathways. SLAMF6 CD2 AIRE IFIT3 NLRP1 HLA-DRB5 TNFRSF18 CCL3 ITGAL BBS1 CD96 CD320 CD19 Journal of Oncology 9 +e melanoma patients were first classified according to the median protection value as the cut-off. Compared with the low-risk patients, the high-risk patients are associated with a status of BRAF and NRAS mutation. In the GEO cohort, the high-risk patients had a significantly shorter OS than the shorter OS (High-risk vs. low-risk: median OS, 1.66 vs. 3.82 years; HR � 3.14, 95% confidence interval [CI] 2.07 to 4.78; low-risk patients (Figures 4(a)–4(c)), except for the NRAS p< 0.0001; Figure 2(b)). To validate the prognosis prediction mutation cohort (High-risk vs. low-risk: median OS, 3.34 vs. of the 8-gene-based gene signature, we calculated the pro- 3.57 years; HR � 1.82, 95% CI 0.69 to 4.81; p � 0.2267; tection value of each patient in the TCGA database using the Figure 4(d)). Subsequently, we validated these new findings same formula. Similarly, the patients were classified into using the TCGA database. Similarly, patients with BRAF and high- and low-risk groups using the same method. Ex- NRAS status were selected to validate the local gene ex- pectedly, the OS in the high-risk group was shorter than that pression patterns, and the OS in the high-risk group was in the low-risk group (High-risk vs. low-risk: median OS, shorter than that in the low-risk group in all cohorts (Figures 4(e), 4(f) and Supplementary Figures 5(a) and 1.66 vs. 3.82 years; HR � 1.75, 95% CI 1.33 to 2.30; p< 0.0001; Figure 2(c)). 5(b)). Taken together, the gene risk signature-based classi- fication could accurately identify patients with poor prog- nosis regardless of the BRAF and NRAS status. 3.3. Correlations between the Local Gene Signature and Prognostic Features in Melanoma. +e baseline character- istics of the GEO and TCGA cohort were compared based on 3.5. Low-Risk Melanoma Patients Exhibited an Enhanced the local gene signature, and the comparisons are shown in Local Immune Phenotype. Considering the distinct prog- Supplementary Tables 7 and 8. Overall, in the TCGA cohort, nosis based on gene signature, we explored the phenotypical the age at diagnosis (High-risk vs. low-risk: mean age, 55.7 differences between the risk groups using genome expres- vs. 60.5 years, p � 0.002), advanced AJCC (American Joint sion data. Melanoma is known as the most common im- Committee on Cancer) stage (High-risk vs. low-risk: rate, mune-related malignancy, and melanoma patients were the 55.3% vs. 35.7%, p< 0.001), and NRAS mutation rate (High- first to benefit from immunotherapy. Hence, the five most risk vs. low-risk: rate, 32.2% vs. 23.1%, p � 0.037) signifi- common immune-related gene sets (adaptive immune re- cantly differed between the high- and low-risk groups sponse M13847, activation of immune response M19789, (Supplementary Table 7). +e GEO cohort exhibited a activation of immune innate response M15340, adaptive similar distribution (Supplementary Table 8). Next, we immune response based on immune receptors M11342, and performed univariate and multivariate Cox regression an- primary immunodeficiency syndrome M7603) were alyses using the TGGA database and revealed that the local extracted from the Molecular Signatures Database, and an gene-related risk signature was independently correlated immune-related gene set was created. Interestingly, com- with OS (Table 1). Furthermore, the local gene-related risk pared with the high-risk group, the GSEA revealed a highly signature was validated as an independent factor using the significant enrichment of immune-related phenotypes in the GEO database (Supplementary Table 9), confirming that this low-risk group (Figures 5(a) and 5(b)), indicating that pa- signature independently predicts prognosis with strong tients with the low-risk gene signature had an intense local power. immune response microenvironment. Next, the patients in In addition, as shown by the Kaplan–Meier OS curves the GEO database were divided into high- and low-risk (Figure 3(a)), the OS during early-stage melanoma signifi- groups according to their protection values. As presented in cantly differed between the high- and low-risk groups of Figure 5(c), the genes forming the gene risk signature patients in the GEO datasets (High-risk vs. low-risk: median exhibited distinct expression patterns that corresponded to OS, 2.15 vs. 4.84 years; HR � 3.50, 95% CI 2.00 to 6.13; the protection value. +e low-risk patients exhibited high p< 0.0001). Additionally, although not statistically signifi- expression levels of T cell activation-related genes cant, a divergence appeared to emerge in the OS curves prior (TNFSF14, AIRE, CD2, and CD19), NK cell activation-re- to 10 years of follow-up in patients at the advanced-stage lated genes (SLAMF6 and NKTR), and autoimmune-related (High-risk vs. low-risk: median OS, 1.38 vs. 2.79 years; genes (Figure 5(c)). Additionally, the low-risk patients had HR � 1.75, 95% CI 0.88 to 3.48; p � 0.1081; Figure 3(b)). higher expression levels of a crucial negative regulator of the Subsequently, we validated our novel findings using the immune system (CTLA4) and a protective gene TCGA database. During both the early (High-risk vs. low- (TNFRSF10B). However, the signature value did not differ risk: median OS, 5.56 vs. 12.61 years; HR � 1.82, 95% CI 1.27 between the cases stratified by age at diagnosis, gender, and to 2.60; p � 0.0012; Figure 3(c)) and advanced (High-risk vs. molecular subtype. Furthermore, the Circos diagram of the low-risk: median OS, 2.86 vs. 5.76 years; HR � 1.78, 95% CI GO analysis illustrates the identical tendency of the im- 1.16 to 2.75; p � 0.0085; Figure 3(d)) stages, the gene sig- mune-related genes between the two groups (Figure 5(d)). nature had prognostic significance. Furthermore, the sig- nature protection value differed between patients stratified 4. Discussion by clinical and AJCC stages (Figures 3(e) and 3(f)). In this study, we firstly identified a gene signature that was 3.4. Application of the Local Gene Signature in Stratified significantly associated with OS in patients with melanoma Melanoma Cohorts. In this study, we evaluated the prog- using gene expression data from the GEO and TCGA da- nostic value of the local gene signature in stratified cohorts. tabases. Furthermore, different immune gene patterns were 10 Journal of Oncology and contributes to malignancy progression and an unfa- observed in high- and low-risk patients. In the patients with the low-risk gene signature, the innate and adaptive immune vorable prognosis [28, 29], while highly expressed TNFSF14 in human melanoma cells and microvesicles may contribute systems are capable of coordinating a robust immune re- sponse, indicating a need for a distinct immunotherapy to the mediation of T cell responses to cancer cells [30]. strategy according to the gene expression pattern. Notably, although the available genomic and associated +e identification of molecular subtypes in other ma- clinical data have been verified, several genes constituting lignancies provided the impetus to utilize transcriptome our signature have not been studied in melanoma. However, profiling to explore the gene expression patterns in mela- these genes appear to exert oncogenic or tumor suppressive noma. Previous studies have used mRNA expression pro- functions in other tumors. For instance, GPR87 plays a critical oncogenic role in pancreatic cancer progression, and filing to distinguish among the subtypes of lymphoma with a high degree of accuracy [19, 20]. Importantly, parallel studies SH3GL3 is a novel invasion-associated candidate gene that likely contributes to the invasive genotype of malignant in melanoma also revealed that patients could be grouped into “molecular subtypes” with very different biological gliomas [31, 32]. Furthermore, our gene risk signature remained an independent prognostic predictor after properties that clinically behave as different disease entities [3, 4, 7, 21, 22]. In the present study, a large sample of adjusting for the clinicopathological and molecular features melanoma cases from the GEO database, including 37 (Table 1). primary and 85 metastatic melanoma cases, was used as a To better understand the gene signature influencing discovery set. In the preliminary analysis, markedly distinct patient survival, we conducted a subgroup analysis and local gene phenotypes were observed based on the clinical mainly focused on tumor stage (clinical or AJCC stage) and stage (primary vs. metastatic), particularly in telomere molecular characteristics (BRAF and NRAS status). In the GEO database, more than 80 metastatic melanomas were maintenance, telomeric DNA binding, biosynthetic process, and metastasis (Figures 1(a), 1(b) and Supplementary analyzed, and this sample size was sufficient to display the power of the gene signature in predicting the outcome even Figures 1–4). Consistent results have been reported in previous studies, in which telomere maintenance, cancer after adjusting for the clinical stage (Figures 3(a) and 3(b)). Furthermore, the clinical implications according to mo- metabolism, and DNA repair were highly associated with malignancy progression, and poor clinical outcomes were lecular grading and staging were immediately validated predominant in melanoma with high malignancy [23–25]. using the TCGA dataset (Figures 3(c) and 3(d)). Moreover, However, differences across the entire gene set have not been as shown in the protection value pattern presented in identified in melanoma patients at different stages. Ac- Figures 3(e) and 3(f), the advanced-stage melanomas un- cordingly, we are the first to demonstrate that the overall derwent a malignant course in our gene expression model. gene expression pattern in melanoma patients is positively In general, the BRAF and NRAS status defined the nature of the proliferative apparatus, which has been well established distinguished by the malignant grade (Figures 1(c) and 1(d)). as a major molecular biomarker of melanoma [33, 34]. +e gene-related risk feature might have contributed to the poor To the best of our knowledge, establishing precise sig- natures to determine the status of patients is refreshing prognosis in the patients regardless of BRAF status in both because these signatures are powerful prognostic predictors the discovery and the validation databases (Figures 4(a), 4(b) and, if correctly applied, can enable patient stratification to and Supplementary Figure 5). Similarly, despite the un- achieve better immunotherapeutic outcomes. Numerous certain results using the GEO dataset (Figures 4(c) and 4(d)), studies have investigated both single prognostic biomarkers a Kaplan–Meier analysis of TCGA patients suggested that and local immune parameters in patients with melanoma patients with the high-risk gene signature had a worse OS [26, 27]. However, the prognostic value of systemic gene than patients with the low-risk signature in both the NRAS signatures remains unclear. In our study, we identified two wild-type and mutant subgroups (Figures 4(e) and 4(f)). Considering that BRAF and NRAS mutant melanomas are gene expression patterns and generated an 8-gene-based (GPR87, KIT, SH3GL3, PVRL1, ATP1B1, CDAN1, FAU, prone to an ominous prognostic outcome [35, 36], the similar gene expression signature pattern of the wild-type and TNFSF14) gene signature that could recognize mela- noma patients with a high risk of unfavorable clinical and mutant melanomas indicates that our model-based outcomes. Next, we tested the signature using the GEO classification could accurately identify patients with unfa- database (for discovery) and validated the signature using vorable prognoses regardless of the BRAF and NRAS status. the TCGA database (Figures 2(a)–2(c)). Our signature However, the exact mechanism remains unknown and consists of diverse genes comprising protective (ATP1B1, should be further examined. CDAN1, FAU, and TNFSF14)) and risky (GPR87, KIT, Data from previous studies investigating the response of melanoma to immune checkpoint inhibitors have illustrated SH3GL3, and PVRL1), which could be considered gene- related protective and risk patterns in melanoma. Alto- the need to develop a strategy to consider stratification based on the gene signature. Due to its formation of related genes, gether, our findings may prompt a novel treatment strategy to improve prognosis by shaping the gene signature. the signature was highly associated with the overall intensity of the local immune response. +e designated low-risk According to previous studies, the genes forming our sig- nature could be considered promising therapeutic targets patients exhibited an enhanced local immune phenotype due to their nature and prognostic impact. KIT, which is a compared to the low-risk patients (Figures 5(a) and 5(b)). famous oncogene, is often mutated in advanced melanoma Interestingly, the local immune signature pattern was Journal of Oncology 11 compatible with prognosis determination in patients at a Authors’ Contributions low- or high-risk, clearly suggesting that high-risk patients Jing Wang, Peng-Fei Kong, and Hai-Yun Wang equally share similar decreased immune abilities in determining contributed to this work. prognosis, even with the different intensity of T cell acti- vation-related genes (TNFSF14, AIRE, CD2, and CD19), NK cell activation-related genes (SLAMF6, and NKTR), and Acknowledgments autoimmune-related genes (HLA-DOB, HLA-DOA, IL7R, and TNFRSF18) (Figures 5(c) and 5(d)). However, this result +e authors thank Dr. Hao-Tu Zhu for the statistical advice is consistent with those of previous reports showing that the and reviewing the manuscript. +is study was partially immune response against tumors increased the survival time supported by the National Natural Science Foundation of of patients with advanced-stage tumors, such as melanoma, China (81602468) and Natural Science Foundation of Anhui lung cancer, and hepatocellular carcinomas [37–39]. In (1808085MH286). addition, low-risk patients have higher expression levels of a crucial negative regulator of the immune system (CTLA4) Supplementary Materials and a protective gene (TNFRSF10B). According to pre- liminary reports, melanoma patients can benefit from anti- Supplementary Table 1: the characteristics of patients from CTLA treatment [40, 41], highlighting the potential of our the GEO database. Supplementary Table 2: the character- signature to identify melanoma patients in which the use of istics of patients from the TCGA database. Supplementary immune checkpoint inhibitors is effective. Table 3: comparison of clinicopathologic characteristics Additionally, several limitations of this study must be between primary and metastatic melanoma in GEO data- addressed. First, our study is limited since it is retrospective base. Supplementary Table 4: biological processes enriched and should be validated by prospective studies. Second, to in the metastatic group. Supplementary Table 5: biological achieve better clinical application, the validity of our sig- processes enriched in the primary group. Supplementary nature in predicting responses to immunotherapy and re- Table 6: fourteen genes with prognostic value in CEO lationship with hyperprogressive disease (HPD) should be melanoma patients. Supplementary Table 7: comparison of tested; as previously reported, local immune factors are clinicopathologic characteristics between high and low-risk potentially involved in the formation of this signature [42]. melanoma in the TCGA database. Supplementary Table 8: Finally, functional and mechanistic studies should be per- comparison of clinicopathologic characteristics between formed to investigate the 8 genes alone or in combination to high and low-risk melanoma in the GEO database. Sup- support the clinical application of our signature. plementary Table 9: multivariate Cox regression analysis of clinicopathologic factors for overall survival in the GEO database. Supplementary Figure 1: gene set enrichment 5. Conclusions analysis (GSEA) for comparing genotype between metastatic and primary. FDR � false discovery rate; NES � normalized We identified an 8-gene signature (i.e., GPR87, KIT, enrichment score. Supplementary Figure 2: gene set en- SH3GL3, PVRL1, ATP1B1, CDAN1, FAU, and TNFSF14) richment analysis (GSEA) for comparing genotype between with independent prognostic value on melanoma. Addi- metastatic and primary. FDR � false discovery rate; tionally, the gene expression pattern correlated with mela- NES � normalized enrichment score Supplementary Fig- noma-immune microenvironment and immune-related ure 3: gene set enrichment analysis (GSEA) for comparing therapy. genotype between metastatic and primary. FDR � false discovery rate; NES � normalized enrichment score. Sup- Abbreviations plementary Figure 4: gene set enrichment analysis (GSEA) for comparing genotype between metastatic and primary. GEO: Gene Expression Omnibus TCGA: +e Cancer Genome Atlas FDR � false discovery rate; NES � normalized enrichment score. Supplementary Figure 5: (a) survival curves of overall PCA: Principal components analysis GSEA: Gene set enrichment analysis, GO gene oncology survival for high and low-risk groups classified by the local gene signature in BRAF wild-type melanoma patients analysis NES: +e normalized enrichment score (TCGA database); (b) survival curves of overall survival for high and low-risk groups classified by the local gene sig- FDR: False discovery rate. nature in BRAF mutation melanoma patients (TCGA da- tabase). (Supplementary Materials) Data Availability All the data supporting the conclusions of this article are References included in the article and its supplementary information files. [1] R. D. Zhang, J. E. Price, G. Schackert, K. Itoh, and I. J. Fidler, “Malignant potential of cells isolated from lymph node or brain metastases of melanoma patients and implications for Conflicts of Interest prognosis,” Cancer Research, vol. 51, no. 8, pp. 2029–2035, +e authors have no conflicts of interest to declare. 1991. 12 Journal of Oncology [2] B. P. Jain, “Prognostic model for primary melanoma,” Annals [19] A. A. Alizadeh, M. B. Eisen, R. E. Davis et al., “Distinct types of of Internal Medicine, vol. 126, no. 10, p. 832, 1997. diffuse large B-cell lymphoma identified by gene expression [3] K. Harbst, J. Staaf, M. Lauss et al., “Molecular profiling reveals profiling,” Nature, vol. 403, no. 6769, pp. 503–511, 2000. low- and high-grade forms of primary melanoma,” Clinical [20] T. R. Golub, D. K. Slonim, P. Tamayo et al., “Molecular Cancer Research, vol. 18, no. 15, pp. 4026–4036, 2012. classification of cancer: class discovery and class prediction by [4] G. Jonsson, C. Busch, S. Knappskog et al., “Gene expression gene expression monitoring,” Science, vol. 286, no. 5439, profiling-based identification of molecular subtypes in stage pp. 531–537, 1999. IV melanomas with different clinical outcome,” Clinical [21] C. April, B. Klotzle, T. Royce et al., “Whole-genome gene expression profiling of formalin-fixed, paraffin-embedded Cancer Research, vol. 16, no. 13, pp. 3356–3367, 2010. [5] C. A. Barker, M. A. Postow, S. A. Khan et al., “Concurrent tissue samples,” PLoS One, vol. 4, no. 12, Article ID e8162, radiotherapy and ipilimumab immunotherapy for patients 2009. with melanoma,” Cancer Immunology Research, vol. 1, no. 2, [22] A. A. Tarhini, Y. Lin, H.-M. Lin et al., “Expression profiles of pp. 92–98, 2013. immune-related genes are associated with neoadjuvant ipi- [6] D. Black and L. Brockway-Lunardi, “+e melanoma research limumab clinical benefit,” Oncoimmunology, vol. 6, no. 2, alliance: the power of patient advocacy to accelerate research Article ID e1231291, 2017. and novel therapies,” Cancer Immunology Research, vol. 1, [23] E. Gaude and C. Frezza, “Tissue-specific and convergent no. 6, pp. 357–361, 2013. metabolic transformation of cancer correlates with metastatic potential and patient survival,” Nature Communications, [7] W. Hugo, J. M. Zaretsky, L. Sun et al., “Genomic and tran- scriptomic features of response to anti-PD-1 therapy in vol. 7, p. 13041, 2016. metastatic melanoma,” Cell, vol. 165, no. 1, pp. 35–44, 2016. [24] K. J. Ransohoff, W. Wu, H. G. Cho et al., “Two-stage genome- [8] K. Pilipow, A. Roberto, M. Roederer, T. A. Waldmann, wide association study identifies a novel susceptibility locus D. Mavilio, and E. Lugli, “IL15 and T-cell stemness in T-cell- associated with melanoma,” Oncotarget, vol. 8, no. 11, based cancer immunotherapy,” Cancer Research, vol. 75, pp. 17586–17592, 2017. no. 24, pp. 5187–5193, 2015. [25] N. Viceconte, M.-S. Dheur, E. Majerova et al., “Highly ag- [9] Z. Wang, C. D. Monteiro, K. M. Jagodnik et al., “Extraction gressive metastatic melanoma cells unable to maintain telo- and analysis of signatures from the gene expression Omnibus mere length,” Cell Reports, vol. 19, no. 12, pp. 2529–2543, by the crowd,” Nature Communications, vol. 7, no. 1, p.12846, 2016. [26] F. Donskov, “Immunomonitoring and prognostic relevance [10] A. Chatterjee, P. A. Stockwell, E. J. Rodger, M. F. Parry, and of neutrophils in clinical trials,” Seminars in Cancer Biology, M. R. Eccles, “scan_tcga tools for integrated epigenomic and vol. 23, no. 3, pp. 200–207, 2013. transcriptomic analysis of tumor subgroups,” Epigenomics, [27] J. J. Luke, K. T. Flaherty, A. Ribas, and G. V. Long, “Targeted vol. 8, no. 10, pp. 1315–1330, 2016. agents and immunotherapies: optimizing outcomes in mel- [11] S. K. Ratuapli, A. E. Bharucha, J. Noelting, D. M. Harvey, and anoma,” Nature Reviews Clinical Oncology, vol. 14, no. 8, A. R. Zinsmeister, “Phenotypic identification and classifica- pp. 463–482, 2017. tion of functional defecatory disorders using high-resolution [28] C. M. Lovly, K. B. Dahlman, L. E. Fohn et al., “Routine anorectal manometry,” Gastroenterology, vol. 144, no. 2, multiplex mutational profiling of melanomas enables en- pp. 314–322, 2013. rollment in genotype-driven therapeutic trials,” PLoS One, [12] A. G. Huth, S. Nishimoto, A. T. Vu, and J. L. Gallant, “A vol. 7, no. 4, Article ID e35309, 2012. continuous semantic space describes the representation of [29] K. T. Montone, P. van Belle, R. Elenitsas, and D. E. Elder, thousands of object and action categories across the human “Proto-oncogene c-kit expression in malignant melanoma: brain,” Neuron, vol. 76, no. 6, pp. 1210–1224, 2012. protein loss with tumor progression,” Modern Pathology, [13] W. Cheng, X. Ren, C. Zhang et al., “Bioinformatic profiling vol. 10, no. 9, pp. 939–944, 1997. identifies an immune-related risk signature for glioblastoma,” [30] R. Mortarini, A. Scarito, D. Nonaka et al., “Constitutive ex- Neurology, vol. 86, no. 24, pp. 2226–2234, 2016. pression and costimulatory function of LIGHT/TNFSF14 on [14] G. Dennis Jr., B. T. Sherman, D. A. Hosack et al., “DAVID: human melanoma cells and melanoma-derived micro- Database for Annotation, Visualization, and Integrated Dis- vesicles,” Cancer Research, vol. 65, no. 8, pp. 3428–3436, 2005. [31] S. Delic, N. Lottmann, K. Jetschke, G. Reifenberger, and covery,” Genome Biology, vol. 4, no. 9, 2003. [15] M. A. Harris, J. Clark, A Ireland et al., “+e Gene Ontology M. J. Riemenschneider, “Identification and functional vali- (GO) database and informatics resource,” Nucleic Acids Re- dation of CDH11, PCSK6 and SH3GL3 as novel glioma in- search, vol. 32, 32, no. 90001, pp. D258–D261, 2004. vasion-associated candidate genes,” Neuropathology and [16] J. Yao, O. L. Caballero, Y. Huang et al., “Altered expression Applied Neurobiology, vol. 38, no. 2, pp. 201–212, 2012. and splicing of ESRP1 in malignant melanoma correlates with [32] L. Wang, W. Zhou, Y. Zhong et al., “Overexpression of G epithelial-mesenchymal status and tumor-associated immune protein-coupled receptor GPR87 promotes pancreatic cancer cytolytic activity,” Cancer Immunology Research, vol. 4, no. 6, aggressiveness and activates NF-kappaB signaling pathway,” pp. 552–561, 2016. Molecular Cancer, vol. 16, no. 1, p. 61, 2017. [17] I. S. Lossos, D. K. Czerwinski, A. A. Alizadeh et al., “Prediction [33] S. M. C. Broekaert, R. Roy, I. Okamoto et al., “Genetic and of survival in diffuse large-B-cell lymphoma based on the morphologic features for melanoma classification,” Pigment expression of six genes,” New England Journal of Medicine, Cell & Melanoma Research, vol. 23, no. 6, pp. 763–770, 2010. vol. 350, no. 18, pp. 1828–1837, 2004. [34] G. V. Long, A. M. Menzies, A. M. Nagrial et al., “Prognostic [18] A. Subramanian, P. Tamayo, V. K. Mootha et al., “Gene set and clinicopathologic associations of oncogenic BRAF in enrichment analysis: a knowledge-based approach for inter- metastatic melanoma,” Journal of Clinical Oncology, vol. 29, preting genome-wide expression profiles,” Proceedings of the no. 10, pp. 1239–1246, 2011. National Academy of Sciences, vol. 102, no. 43, pp. 15545– [35] L. N. Kwong, J. C. Costello, H. Liu et al., “Oncogenic NRAS 15550, 2005. signaling differentially regulates survival and proliferation in Journal of Oncology 13 melanoma,” Nature Medicine, vol. 18, no. 10, pp. 1503–1510, [36] A. Viros, J. Fridlyand, J. Bauer et al., “Improving melanoma classification by integrating genetic and morphologic fea- tures,” PLoS Medicine, vol. 5, no. 6, p. e120, 2008. [37] D. Sia, Y. Jiao, I. Martinez-Quetglas et al., “Identification of an immune-specific class of hepatocellular carcinoma, based on molecular features,” Gastroenterology, vol. 153, no. 3, pp. 812–826, 2017. [38] B. A. Weir, M. S. Woo, G. Getz et al., “Characterizing the cancer genome in lung adenocarcinoma,” Nature, vol. 450, no. 7171, pp. 893–898, 2007. [39] K. Yoshihara, M. Shahmoradgoli, E. Martinez et al., “Inferring tumour purity and stromal and immune cell admixture from expression data,” Nature Communications, vol. 4, p. 2612, [40] Trial watch: Ipilimumab success in melanoma provides boost for cancer immunotherapy,” Nature Reviews Drug Discovery, vol. 9, no. 8, pp. 584-585, 2010. [41] E. M. Van Allen, D. Miao, B. Schilling et al., “Genomic correlates of response to CTLA-4 blockade in metastatic melanoma,” Science, vol. 350, no. 6257, pp. 207–211, 2015. [42] S. Champiat, L. Dercle, S. Ammari et al., “Hyperprogressive disease is a new pattern of progression in cancer patients treated by anti-PD-1/PD-L1,” Clinical Cancer Research, vol. 23, no. 8, pp. 1920–1928, 2017.
Journal of Oncology – Hindawi Publishing Corporation
Published: Apr 29, 2020
You can share this free article with as many people as you like with the url below! We hope you enjoy this feature!
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.