Introduction

Oncology Letters

1792-1074 1792-1082

D.A. Spandidos

10.3892/ol.2020.11287

OL-0-0-11287

Articles

Prognostic value of key genes of the JAK-STAT signaling pathway in patients with cutaneous melanoma

Pan

Fuqiang

1 * Wang

Qiaoqi

1 * Li

Sizhu

2 Huang

Rui

3 Wang

Xiangkun

4 Liao

Xiwen

4 Mo

Haiyan

1 Zhang

Liming

1 Zhou

Xiang

1Department of Medical Cosmetology, The Second Affiliated Hospital of Guangxi Medical University, Guangxi Zhuang Autonomous Region 530000, P.R. China 2Department of Medical Oncology, The First Affiliated Hospital of Guangxi Medical University, Guangxi Zhuang Autonomous Region 530000, P.R. China 3Department of Hematology, The First Affiliated Hospital of Guangxi Medical University, Guangxi Zhuang Autonomous Region 530000, P.R. China 4Department of Hepatobiliary Surgery, The First Affiliated Hospital of Guangxi Medical University, Guangxi Zhuang Autonomous Region 530000, P.R. China

Correspondence to: Dr Xiang Zhou, Department of Medical Cosmetology, The Second Affiliated Hospital of Guangxi Medical University, 166 Daxue East Road, Nanning, Guangxi Zhuang Autonomous Region 530000, P.R. China, E-mail: zx_gxmu@hotmail.com *

Contributed equally

03 2020

10 01 2020

19 3 1928 1946 14022019 22112019

2020

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

The Janus kinase-signal transducer and activator of transcription (JAK-STAT) signaling pathway is involved in cell immunity, division and death, as well as in tumor formation. The expression of key genes in the JAK-STAT signaling pathway in different types of cancer serves different roles. However, few reports are available on the prognostic value of the genes of the JAK-STAT signaling pathway in skin cutaneous melanoma (SKCM). The potential prognostic value of gene expression in the JAK-STAT signaling pathway in patients with SKCM was analyzed in the present study using data obtained from The Cancer Genome Atlas. To predict the potential functions and mechanisms of these genes in SKCM, gene set enrichment analysis (GSEA) and bioinformatics analysis were performed. A nomogram model including gene expression level and high risk factors was used to predict the risk level of prognostic. High expression levels of STAT1, STAT3, STAT4 and STAT5B, and low expression levels of STAT6 were associated with favorable prognosis [adjusted P<0.001; hazard ratio (HR), 0.595; 95% confidence interval (CI), 0.455–0.778; adjusted P=0.018; HR, 0.725; 95% CI, 0.555–0.947; adjusted P<0.001; HR, 0.590; 95% CI, 0.450–0.773; adjusted P=0.007; HR, 0.690; 95% CI, 0.526–0.940; and adjusted P=0.026; HR, 0.737, 95% CI, 0.563–0.964, respectively]. GSEA results demonstrated that these genes were involved in cell differentiation, invasion, adhesion, migration, cycle, colony formation and mitogen-activated protein kinase signaling. The combination of genes with favorable prognosis had a better effect on the overall survival (univariate survival analysis, P<0.05). The results of the present study suggest that STAT1, STAT3, STAT4, STAT5B and STAT6 gene expression may be used as a potential prognostic biomarker of SKCM, and the combined outcomes may exhibit a stronger interaction and higher survival time for SKCM.

melanoma Janus kinase-signal transducer and activator of transcription signaling pathway prognosis

Introduction

Skin cutaneous melanoma (SKCM) is a highly aggressive skin cancer, which arises from the malignant transformation of melanocytes in the basal layer of the epidermis (1,2) and has a poor prognosis, with a 5-year overall survival (OS) of 91.8% worldwide (3). The number of new cases of SKCM that will emerge in 2019 was estimated to be 96,480, with the mortality estimated to be 7,230 (4,5). Therefore, there is an urgent need to identify novel biomarkers and prognostic predictive indicators for the detection and management of SKCM.

The JAK-STAT signaling pathway serves a crucial role in cell immunity, division and death, and in tumor formation (6). The two key components involved in this pathway are Janus kinases (JAKs) and signal transducer and activator of transcription proteins (STATs), which are encoded by the genes JAK (JAK1, JAK2, JAK3 and TYK2) and STAT (STAT1, STAT2, STAT3, STAT4, STAT5A, STAT5B and STAT6), respectively (6).

A previous study has demonstrated that the key genes involved in the JAK-STAT signaling pathway are associated with several types of cancer, including breast (7,8), ovarian, lung, brain (9) and colorectal (10) cancer, and that their differential expression may result in different prognosis outcomes in different types of cancer. However, few reports about the association between these genes and SKCM are available, and further investigation analyzing the prognostic value of gene expression in the JAK-STAT signaling pathway in SKCM is needed. The aim of the present study was to identify the prognostic values of the expression of genes involved in the JAK-STAT signaling pathway in patients with SKCM based on data derived from public databases and bioinformatics analysis, and to explore the underlying mechanism that may affect the outcome in SKCM prognosis.

Materials and methods <sec> <title>Data preparation

The Cancer Genome Atlas (TCGA; https://cancergenome.nih.gov; accessed on September 10th 2018 was used to obtain the gene expression and clinical data of patients with SKCM, including sex, age and tumor stage. A total of 458 SKCM cases were selected after removing the cases with missing mRNA expression or clinical data and 0-day survival time.

Functional analysis of key genes in the JAK-STAT signaling pathway

The Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway map was generated using the KEGG website (https://www.kegg.jp; accessed on September 10th 2018 (11–13). Gene ontology (GO) term analysis, including biological function (BP), molecular function (MF) and cellular component (CC), as well as KEGG enrichment analysis for JAK and STAT gene families, were performed using the Database for Annotation, Visualization and Integrated Discovery version 6.8 (DAVID; https://david.ncifcrf.gov/tools.jsp; accessed on September 13th 2018. The official gene symbol was used as the identifier; the species was Homo sapiens (14,15).

Gene interaction and association analysis

Gene-gene interaction analysis was performed using gene multiple association network integration algorithm (GeneMANIA; http://genemania.org; accessed on September 15th 2018 using the default parameters (16,17). The correlation between JAK and STAT pathway gene expression in SKCM was evaluated using the Pearson's correlation coefficient; P<0.01 was considered to indicate a significant correlation. Protein-protein interaction analysis was performed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; https://string-db.org; accessed on September 25th 2018. The minimum required interaction score was 0.400 (18,19).

Patient grouping based on gene expression level

Patients with SKCM were divided into high or low expression groups using the median value, and the vertical scatter plots were generated.

Survival analysis

Kaplan-Meier survival plots with log-rank test were used to evaluate the OS for the high and low-expression groups of each gene and clinicopathological characteristics. In addition, the Cox proportional hazard regression model was used for univariate and multivariate survival analyses, and 95% confidence intervals (CIs) and hazard ratios (HRs) were calculated. Joint effects analysis was performed for the combination of genes identified as significant by the survival analysis.

Nomogram

A nomogram was used to evaluate the association between JAK and STAT gene expression and clinical information in SKCM. In addition, the risk rank for each gene and clinicopathological characteristic, including age, sex and tumor stage, was evaluated by the risk points and total points. Survival rates (1-, 3- and 5-year) were also scored. A high score was associated with a low survival rate.

Gene set enrichment analysis (GSEA)

To investigate the potential underlying mechanism of the differential expression of STAT1, STAT3, STAT4, STAT5B and STAT6 in SKCM, GSEA version 3.0 software (http://software.broadinstitute.org/gsea/msigdb/index.jsp; accessed on September 20th 2018 (20) was used, and the difference in the expression levels in the low and high-expression groups for each gene was analyzed with reference gene sets, which were based on the Molecular Signatures Database sets c2 (KEGG gene sets, c2.all.v6.2.symbols.gmt), c5 [Gene Ontology (GO) gene sets, c5.all.v6.2.symbols.gmt] and c6 (oncogenic signatures gene sets, c6.all.v6.2.symbols.gmt) (21). The permutation number was set to 1,000. Those enrichment gene sets revealed by GSEA as exhibiting a nominal P<0.05 and a false discovery rate (FDR) <0.25 were considered to indicate a statistically significant difference. The default parameters were used in GSEA software.

Statistical analysis

Statistical analysis was performed using SPSS v.25.0 software (IBM Corp.) and R 3.3.5 (https://www.r-project.org/). Kaplan-Meier survival analysis and the log-rank test were used to calculate the OS and P-values for all associations. The Cox proportional hazards regression model was used for univariate and multivariate survival analyses. HRs and 95% CIs were calculated using the Cox proportional hazards regression model with adjustment for influential clinical characteristics such as race, sex, age, Tumor-Node-Metastasis stage (22) and body mass index. FDRs in the GSEA were adjusted for multiple testing with the Benjamini-Hochberg procedure to control the FDR (23,24). P<0.05 was considered to indicate a statistically significant difference. Vertical scatter plots and survival curves were generated in GraphPad Prism v.7.0 (GraphPad Software, Inc.).

Results <sec> <title>Clinicopathological characteristics of patients with SKCM

The clinicopathological data of the 458 patients included in the present study are presented in Table I. Race, age and tumor stage were significantly associated with median survival time (P=0.003, P<0.001 and P<0.001, respectively; Table I). After normalization, only JAK1, STAT1, STAT3, STAT4, STAT5B and STAT6 genes mRNA expression were observed.

Bioinformatics analysis of JAK and STAT genes

The KEGG pathway map of the JAK-STAT signaling pathway is represented in Fig. 1 (KEGG map no. 04630; http://www.genome.jp/dbget-bin/www_bget?pathway:map04630). The GO term and KEGG enrichment analysis for JAK and STAT genes included BP (Figs. S1A and S2A), CC (Figs. S1B and S2B), MF (Figs. S1C and S2C) and KEGG (Figs. S1D and S2D). The DAVID result for the combination of JAK and STAT gene families suggested that the JAK and STAT signaling pathways were associated with cell migration, the mitogen-activated protein kinase (MAPK) cascade, cell differentiation (Fig. 2A), cytosol, cytoplasm (Fig. 2B), DNA binding, ATP binding, signal transducer activity (Fig. 2C), the PI3K-AKT signaling pathway and the pancreatic cancer signaling pathway (Fig. 2D).

Gene interaction and association analysis

The gene-gene interaction network was analyzed separately in the JAK family (Fig. S3) and in the STAT family (Fig. S4), as well as in their combination (Fig. 3A). The protein-protein interaction network is presented in Fig. 3B.

Pearson's correlation coefficient was used to analyze the correlation between JAK and STAT genes in SKCM tissues based on the TCGA dataset (Fig. 3C). The results suggested that, with the exception of STAT6, the expression of STAT genes (STAT1, STAT3, STAT4 and STAT5B) in SKCM strongly correlated with each other.

Scatter plots of the expression of all JAK and STAT genes in TCGA dataset (using the median as the cut-off value) are presented in Fig. 3D. The difference between the high and low-expression groups was statistically significant (P<0.001).

Survival analysis

The results of univariate survival analysis of JAK and STAT genes are presented in Fig. 4 and Table II. High expression levels of STAT1, STAT3, STAT4 and STAT5B, and low expression levels of STAT6 were associated with a favorable prognosis (P<0.05). Multivariate Cox proportional hazards regression analysis identified that sex, race, age and tumor stage were associated with the prognosis of patients with SKCM. Multivariate survival analysis, in agreement with univariate survival analysis, demonstrated that high expression of STAT1, STAT3, STAT4 and STAT5B, and low expression of STAT6 was associated with a favorable prognosis (adjusted P<0.001; HR, 0.595; 95% CI, 0.455–0.778; adjusted P=0.018; HR, 0.725; 95% CI, 0.555–0.947; adjusted P<0.001; HR, 0.590; 95% CI, 0.450–0.773; adjusted P=0.007; HR, 0.690; 95% CI, 0.526–0.940; and adjusted P=0.026; HR, 0.737; 95% CI, 0.563–0.964, respectively; Table II).

Predictive nomogram and joint effects analysis

Independent factors, including age, sex, tumor stage and mRNA expression, were integrated into the prognostic nomogram to predict the clinical outcomes of patients with SKCM (Fig. 5A). Age, tumor stage, and JAK1, STAT1, STAT4, STAT5B and STAT6 expression levels exhibited major contributions as prognostic signatures in the risk scores (range, 0–100).

All survival related genes, including STAT1, STAT3, STAT4, STAT5B and STAT6, were selected and grouped to perform joint effects analysis. The grouping information of joint effects analysis is presented in Tables III–VI. The group number with simple Arabic numerals represented all gene selected groups (Table III); Arab numerals with brackets are 4 selected genes (Table IV); Lowercase Roman numerals are 3 selected genes (Table V); Capitalized Roman numerals are 2 selected genes (Table VI). The combination of genes with favorable prognosis had a better effect on the OS (univariate survival analysis; P<0.05; Table VII and Figs. 5B-G, 6A-J and 7A-J).

GSEA

The detailed GSEA results, including KEGG, GO and oncogenic signatures, are shown in Tables SI, SII and SIII, respectively, and in Fig. 8. High expression of STAT1 was significantly associated with immune response (normalized P=0.002; FDR=0.243; Fig. 8A), cell adhesion (normalized P=0.002; FDR=0.206; Fig. 8B; normalized P=0.010; FDR=0.241; Fig. 8D) and WNT protein binging (normalized P=0.006; FDR=0.261; Fig. 8C). By contrast, low expression of STAT4 was associated with WNT protein binding (normalized P=0.001; FDR=−0.234; Fig. 8E), whereas high expression of STAT5B was associated with gene silencing (normalized P=0.002; FDR=0.249; Fig. 8F).

Discussion

In the Surveillance, Epidemiology, and End Results Program database, the 5-year relative survival is 98% for prostate cancer, 89.9% for breast cancer and 19.4% for lung cancer (3,4). For melanoma, a 91.8% 5-year relative survival rate appears satisfactory; however, the 5-year relative survival for patients with tumor stage IV is only 3% (22,25).

The JAK-STAT signaling pathway serves a crucial role in functions such as cell proliferation, differentiation, migration and apoptosis, cell survival in hematopoiesis, immune cell development, stem cell maintenance and organismal growth processes (26–28). Dysfunction in the JAK-STAT signaling pathway is associated with diseases such as cancer and immune disorders (6,27). In the JAK-STAT signaling pathway, the JAK family comprises JAK1, JAK2, JAK3 and TYK2, and the STAT family comprises STAT1, STAT2, STAT3, STAT4, STAT5A, STAT5B and STAT6 (6,29). A number of diseases are associated with the expression levels of genes of the JAK-STAT signaling pathway. A previous study has demonstrated that increased expression of STAT3 or STAT1, but not of both, was present in melanoma compared with that in benign nevi or skin melanocytes (30). In addition, active STAT3 was present in melanoma (30). By contrast, overexpression of microRNA (miR)-214 in a lung cancer cell line reduced the expression of JAK1, which inhibited cell proliferation and colony formation, and suppressed cell migration and invasion (31). In papillary thyroid cancer, downregulation of JAK1 by miR-520a-3p inactivated the JAK-STAT signaling pathway (32).

However, studies that explored the association between the prognosis of melanoma and the JAK-STAT signaling pathway are limited. In the present study, the expression of JAK and STAT family genes in melanoma was investigated based on TCGA data. The observation that JAK-STAT gene expression is associated with the MAPK signaling pathway is in agreement with a previous study, which demonstrated that the JAK-STAT signaling pathway is integrated with the MAPK signaling pathway (33–35) and is associated with melanoma (36). Expression of JAK2, JAK3, TYK2, STAT2 and STAT5B was not observed after normalizing mRNA expression in the present study, suggesting that these genes were expressed at a low level. High expression of STAT1, STAT3, STAT4 and STAT5B, as well as low expression of STAT6, were associated with a favorable prognosis of patients with SKCM. These results are consistent with other studies, in which high expression of STAT1 was associated with favorable prognosis in high-grade serous ovarian cancer (HGSC) (37), colorectal cancer (38) and esophageal squamous cell carcinoma (39). In addition, high expression of STAT1 in HGSC was significantly associated with the recruitment of intraepithelial CD8⁺ T cells, which enhanced the prognostic and predictive value of intratumoral CD8⁺ T cells in HGSC (40), potentially due to tumors with high STAT1 mRNA expression exhibiting elevated expression of genes specific for tumor-associated macrophages and immunosuppressive T lymphocytes (7). The results of the enrichment analysis in the present study also revealed that STAT1 was associated with the immune response, which suggested that STAT1 may accelerate the immune cell response to cancer (41). However, high STAT1 expression was associated with poor prognosis in glioblastoma (42), and breast (7,8), ovarian, lung, blood and brain (9) cancer.

In contrast to that of STAT1, STAT3 expression is downregulated in malignant pleural mesothelioma (43); however, the prognostic value of this association has not been reported. The majority of studies on STAT3 and cancer prognosis have demonstrated that phosphorylated STAT3 is associated with a poor outcome in colorectal cancer (10), multiple myeloma (44) and urothelial carcinoma (45). In addition, STAT4 has been detected in several types of cancer, including prostate (46), breast (47), gastric (48) and ovarian (49) cancer. The upregulation of STAT4 in hepatocellular carcinoma is associated with favorable prognosis, possibly due to the expression of STAT4 in the immune cells; however, the function and mechanism of STAT4 in non-immune cells remains unknown (50). High expression of STAT5B in astrocytoma is associated with poor prognosis (51), whereas high expression of STAT6 is associated with poor prognosis in colon cancer (52).

The potential effects of these associations require further exploration. For instance, colorectal carcinoma cell lines exhibiting low STAT1 and high STAT3 expression levels are associated with enhanced tumor growth in xenografts; by contrast, xenograft cell lines with high STAT1 and low STAT3 levels grew slowly (53). Thus, gene interactions may influence the cancer outcome. The different expression level of these two genes in different types of cancer may serve different roles. Joint effects analysis in the present study suggested that any combination of the tested markers may have a higher prognostic value compared with that of any individual biomarker.

The STAT gene family affects cell differentiation (54), invasion (55–57), adhesion (58) and migration (59), as well as cell cycle (60–62) and colony formation (39,63) through the JAK-STAT signaling pathway; these processes are associated with the occurrence, development and outcome of cancer. The potential mechanism of how these genes affect prognosis should be further studied.

The present study investigated the prognostic value of the key genes in the JAK-STAT signaling pathway; only STAT genes were demonstrated to affect the prognosis of SKCM. In melanoma research, no golden standard exists for diagnosis or prognosis that would serve a similar role as the hairy-related 2 gene in breast cancer or prostate specific antigen in prostate cancer. Certain genes may act as biomarkers to predict the prognosis and mechanism of SKCM, including RAF (Raf proto-oncogene), MEK (MAP/ERK kinase), MAPK, RAS, myelocytomatosis oncogene and S100 calcium binding protein (64–69). However, further research is required to identify a golden standard for predicting melanoma.

There were certain limitations in the present study. Firstly, only JAK expression data were reported following normalization; additional mRNA expression data are needed to further confirm these observations. Secondly, this is an association study. Further research is needed to explore the function and mechanism of the genes of the JAK-STAT signaling pathway identified in the present study in patients with SKCM. Another limitation of the current study is the limited sample size. Improved design and larger sample size studies are necessary to validate these results. Finally, there were no expression standards to measure whether the gene expression was high or low.

To the best of our knowledge, the present study is the first to evaluate the association between the expression of genes of the JAK-STAT signaling pathway and OS in patients with SKCM, and to identify the joint effects of prognostic values among the five identified genes. Overall, the results of the present study provided a novel insight into the function of these genes in SKCM clinical outcomes, and may be further utilized in the clinic for predicting the prognosis of SKCM.

In conclusion, the combination of the highly expressed STAT1, STAT3, STAT4 and STAT5B genes, and the lowly expressed STAT6 gene is associated with a favorable prognosis in patients with SKCM, and may be used as a novel biomarker for predicting the prognosis of patients with SKCM. The expression of the genes in the STAT family may affect the prognosis of SKCM by accelerating the immune response and immune cell activity, and by involving the MAPK signaling pathway. Further studies are required to validate these findings.

Supplementary Material

Supporting Data

Acknowledgements

Not applicable.

Funding

The present study was funded by the National Nature Science Foundation of China (grant no. 81760344).

Availability of data and materials

The datasets generated and/or analyzed during the current study are available in the TCGA repository, https://portal.gdc.cancer.gov/.

Authors' contributions

XZ and FP conceived and designed the study. QW, HM and SL processed the data and performed the statistical analysis. LZ, RH, XW and XL wrote and revised the manuscript and helped to perform the analysis and interpretation of data. All authors read and approved the final version of the manuscript.

Ethics approval and consent to participate

The present study did not involve human or animal subjects.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References 1

Goldstein

Diagnosis and management of malignant melanoma

Am Fam Physician63135913682001

11310650

Situm

Buljan

Kolić

Vučić

Melanoma-clinical, dermatoscopical, and histopathological morphological characteristics

Acta Dermatovenerol Croat221122014

24813835

Grimaldi

Simeone

Ascierto

The role of MEK inhibitors in the treatment of metastatic melanoma

Curr Opin Oncol261962032014

10.1097/CCO.0000000000000050

24419498

Ferlay

Soerjomataram

Dikshit

Eser

Mathers

Rebelo

Parkin

Forman

Bray

Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012

Int J Cancer136E359E3862015

10.1002/ijc.29210

25220842

Siegel

Miller

Jemal

Cancer statistics, 2019

CA Cancer J Clin697342019

10.3322/caac.21551

30620402

Aaronson

Horvath

A road map for those who don't know JAK-STAT

Science296165316552002

10.1126/science.1071545

12040185

Tymoszuk

Charoentong

Hackl

Spilka

Müller-Holzner

Trajanoski

Obrist

Revillion

Peyrat

Fiegl

Doppler

High STAT1 mRNA levels but not its tyrosine phosphorylation are associated with macrophage infiltration and bad prognosis in breast cancer

BMC Cancer142572014

10.1186/1471-2407-14-257

24725474

Liu

Huang

Chen

Liu

Wang

Meta-analysis of STAT3 and phospho-STAT3 expression and survival of patients with breast cancer

Oncotarget913060130672018

29560131

Cui

Jing

Long

Tan

Chen

Huang

Xiang

Tian

Zhu

Systematic analysis of gene expression alterations and clinical outcomes of STAT3 in cancer

Oncotarget9319832132018

10.18632/oncotarget.23226

29423040

Han

Sun

Zhao

Zhang

Liu

Zhao

Phosphorylation of STAT3 promotes vasculogenic mimicry by inducing epithelial-to-mesenchymal transition in colorectal cancer

Technol Cancer Res Treat16120912192017

10.1177/1533034617742312

29333928

Kanehisa

Sato

Furumichi

Morishima

Tanabe

New approach for understanding genome variations in KEGG

Nucleic Acids Res47(D1)D590D5952019

10.1093/nar/gky962

30321428

Kanehisa

Furumichi

Tanabe

Sato

Morishima

KEGG: New perspectives on genomes, pathways, diseases and drugs

Nucleic Acids Res45(D1)D353D3612017

10.1093/nar/gkw1092

Kanehisa

Goto

KEGG: Kyoto encyclopedia of genes and genomes

Nucleic Acids Res2827302000

10.1093/nar/28.1.27

10592173

Huang da

Sherman

Lempicki

Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources

Nat Protoc444572009

10.1038/nprot.2008.211

19131956

Huang da

Sherman

Lempicki

Bioinformatics enrichment tools: Paths toward the comprehensive functional analysis of large gene lists

Nucleic Acids Res371132009

10.1093/nar/gkn923

19033363

Mostafavi

Ray

Warde-Farley

Grouios

Morris

GeneMANIA: A real-time multiple association network integration algorithm for predicting gene function

Genome Biol9(Suppl 1)S42008

10.1186/gb-2008-9-s1-s4

18613948

Warde-Farley

Donaldson

Comes

Zuberi

Badrawi

Chao

Franz

Grouios

Kazi

Lopes

The GeneMANIA prediction server: Biological network integration for gene prioritization and predicting gene function

Nucleic Acids Res38W214W2202010

10.1093/nar/gkq537

20576703

von Mering

Huynen

Jaeggi

Schmidt

Bork

Snel

STRING: A database of predicted functional associations between proteins

Nucleic Acids Res312582612003

10.1093/nar/gkg034

12519996

Szklarczyk

Morris

Cook

Kuhn

Wyder

Simonovic

Santos

Doncheva

Roth

Bork

The STRING database in 2017: Quality-controlled protein-protein association networks, made broadly accessible

Nucleic Acids Res45(D1)D362D3682017

10.1093/nar/gkw937

27924014

Subramanian

Tamayo

Mootha

Mukherjee

Ebert

Gillette

Paulovich

Pomeroy

Golub

Lander

Mesirov

Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles

Proc Natl Acad Sci USA10215545155502005

10.1073/pnas.0506580102

16199517

Liberzon

Birger

Thorvaldsdóttir

Ghandi

Mesirov

Tamayo

The molecular signatures database (MSigDB) hallmark gene set collection

Cell Syst14174252015

10.1016/j.cels.2015.12.004

26771021

Balch

Gershenwald

Soong

Thompson

Atkins

Byrd

Buzaid

Cochran

Coit

Ding

Final version of 2009 AJCC melanoma staging and classification

J Clin Oncol27619962062009

10.1200/JCO.2009.23.4799

19917835

Benjamini

Drai

Elmer

Kafkafi

Golani

Controlling the false discovery rate in behavior genetics research

Behav Brain Res1252792842001

10.1016/S0166-4328(01)00297-2

11682119

Reiner

Yekutieli

Benjamini

Identifying differentially expressed genes using false discovery rate controlling procedures

Bioinformatics193683752003

10.1093/bioinformatics/btf877

12584122

Gerami

Cook

Wilkinson

Russell

Dhillon

Amaria

Gonzalez

Lyle

Johnson

Oelschlager

Development of a prognostic genetic signature to predict the metastatic risk associated with cutaneous melanoma

Clin Cancer Res211751832015

10.1158/1078-0432.CCR-13-3316

25564571

Ghoreschi

Laurence

O'Shea

Janus kinases in immune cell signaling

Immunol Rev2282732872009

10.1111/j.1600-065X.2008.00754.x

19290934

Harrison

The Jak/STAT pathway

Cold Spring Harb Perspect Biol4a0112052012

10.1101/cshperspect.a011205

22383755

Zoranovic

Grmai

Bach

Regulation of proliferation, cell competition, and cellular growth by the drosophila JAK-STAT pathway

JAKSTAT2e254082013

24069565

Rane

Reddy

Janus kinases: Components of multiple signaling pathways

Oncogene19566256792000

10.1038/sj.onc.1203925

11114747

Messina

Riker

Munster

Jove

Daud

Activated stat-3 in melanoma

Cancer Control151962012008

10.1177/107327480801500302

18596671

Chen

Jiang

MicroRNA-214 inhibits the proliferation and invasion of lung carcinoma cells by targeting JAK1

Am J Transl Res10116411712018

29736209

Zhang

Guo

MicroRNA-520a-3p suppresses epithelial-mesenchymal transition, invasion, and migration of papillary thyroid carcinoma cells via the JAK1-mediated JAK/STAT signaling pathway

J Cell Physiol23440544672018

10.1002/jcp.27199

30206929

Jain

Zhang

Fong

Lim

Cao

Repression of Stat3 activity by activation of mitogen-activated protein kinase (MAPK)

Oncogene17315731671998

10.1038/sj.onc.1202238

9872331

Rawlings

Rosler

Harrison

The JAK/STAT signaling pathway

J Cell Sci117128112832004

10.1242/jcs.00963

15020666

Nissan

Pratilas

Jones

Ramirez

Won

Liu

Tiwari

Kong

Hanrahan

Yao

Loss of NF1 in cutaneous melanoma is associated with RAS activation and MEK dependence

Cancer Res74234023502014

10.1158/0008-5472.CAN-13-2625

24576830

Davies

Samuels

Analysis of the genome to personalize therapy for melanoma

Oncogene29554555552010

10.1038/onc.2010.323

20697348

Koti

Siu

Clément

Bidarimath

Turashvili

Edwards

Rahimi

Mes-Masson

Squire

A distinct pre-existing inflammatory tumour microenvironment is associated with chemotherapy resistance in high-grade serous epithelial ovarian cancer

Br J Cancer112121512222015

10.1038/bjc.2015.81

25826225

Simpson

Al-Attar

Watson

Scholefield

Ilyas

Durrant

Intratumoral T cell infiltration, MHC class I and STAT1 as biomarkers of good prognosis in colorectal cancer

Gut599269332010

10.1136/gut.2009.194472

20581241

Zhang

Molavi

Lai

The clinical and biological significance of STAT1 in esophageal squamous cell carcinoma

BMC Cancer147912014

10.1186/1471-2407-14-791

25355139

Le Page

Ren

Meunier

Clément

Tyrishkin

Peterson

Kendall-Dupont

Childs

Francis

STAT1-associated intratumoural T_H1 immunity predicts chemotherapy resistance in high-grade serous ovarian cancer

J Pathol Clin Res22592702016

10.1002/cjp2.55

27917296

Chan

Rickert

Vermi

Sheehan

Arthur

Allen

White

Archambault

Lonardi

McDevitt

Dysregulated STAT1-SOCS1 control of JAK2 promotes mammary luminal progenitor cell survival and drives ERalpha(+) tumorigenesis

Cell Death Differ212342462014

10.1038/cdd.2013.116

24037089

Thota

Arimappamagan

Kandavel

Shastry

Pandey

Chandramouli

Hegde

Kondaiah

Santosh

STAT-1 expression is regulated by IGFBP-3 in malignant glioma cells and is a strong predictor of poor survival in patients with glioblastoma

J Neurosurg1213743832014

10.3171/2014.4.JNS131198

24878287

Arzt

Kothmaier

Halbwedl

Quehenberger

Popper

Signal transducer and activator of transcription 1 (STAT1) acts like an oncogene in malignant pleural mesothelioma

Virchows Arch46579882014

10.1007/s00428-014-1584-8

24838635

Jung

Ahn

Choi

Shin

Lee

Yang

Ahn

Kim

Lee

STAT3 expression is associated with poor survival in non-elderly adult patients with newly diagnosed multiple myeloma

Blood Res522932992017

10.5045/br.2017.52.4.293

29333406

Huang

Lee

Chen

Lin

Yeh

Chang

Hsu

Over-expression of activated signal transducer and activator of transcription 3 predicts poor prognosis in upper tract urothelial carcinoma

Int J Med Sci14136013672017

10.7150/ijms.17367

29200949

Lou

Lee

Dhir

DeMiguel

Grandis

Gao

Selective activation of members of the signal transducers and activators of transcription family in prostate carcinoma

J Urol167185918622002

10.1097/00005392-200204000-00095

11912448

Liu

Zhang

Zhao

You

Lin

Zhang

The oncoprotein HBXIP uses two pathways to up-regulate S100A4 in promotion of growth and migration of breast cancer cells

J Biol Chem28730228302392012

10.1074/jbc.M112.343947

22740693

Zhou

Xia

Zhang

Down-regulation of miR-141 induced by helicobacter pylori promotes the invasion of gastric cancer by targeting STAT4

Cell Physiol Biochem33100310122014

10.1159/000358671

24732377

Silver

Naora

Liu

Cheng

Montell

Activated signal transducer and activator of transcription (STAT) 3: Localization in focal adhesions and function in ovarian cancer cell motility

Cancer Res64355035582004

10.1158/0008-5472.CAN-03-3959

15150111

Wang

Chen

Qiang

Wang

Chen

Decreased STAT4 indicates poor prognosis and enhanced cell proliferation in hepatocellular carcinoma

World J Gastroenterol21398339932015

10.3748/wjg.v21.i13.3983

25852285

Kuo

Chen

Tsai

Chai

Kwan

Nucleophosmin overexpression is associated with poor survival in astrocytoma

APMIS1235155222015

10.1111/apm.12381

25907517

Wang

Yuan

Liu

Zhang

Wang

EZH2 and STAT6 expression profiles are correlated with colorectal cancer stage and prognosis

World J Gastroenterol16242124272010

10.3748/wjg.v16.i19.2421

20480530

Nivarthi

Gordziel

Themanns

Kramer

Eberl

Rabe

Schlederer

Rose-John

Knösel

Kenner

The ratio of STAT1 to STAT3 expression is a determinant of colorectal cancer growth

Oncotarget751096511062016

10.18632/oncotarget.9315

27191495

Sun

Han

Wang

Wen

Xia

Xing

Zhang

Correlation of EGFR Del 19 with Fn14/JAK/STAT signaling molecules in non-small cell lung cancer

Oncol Rep36103010402016

10.3892/or.2016.4905

27350337

Jeon

You

Bae

Kim

Nam

Kim

Lee

Dimerization of EGFR and HER2 induces breast cancer cell motility through STAT1-dependent ACTA2 induction

Oncotarget850570505812016

28881584

Chen

Wang

Zhou

Zhu

Hang

Wang

Reversibly stabilized polycation nanoparticles for combination treatment of early- and late-stage metastatic breast cancer

ACS Nano12662066362018

10.1021/acsnano.8b01482

29985577

Merk

Owens

Lopes

Silva

Hussaini

STAT6 expression in glioblastoma promotes invasive growth

BMC Cancer111842011

10.1186/1471-2407-11-184

21595984

Sumiyoshi

Matsushita

Nakamura

Matsuda

Ishiwata

Naito

Uchida

Suppression of STAT5b in pancreatic cancer cells leads to attenuated gemcitabine chemoresistance, adhesion and invasion

Oncol Rep35321632262016

10.3892/or.2016.4727

27035235

Tian

Guan

Zhang

Sun

Zhou

Lin

Ren

Nadeem

Physical interaction of STAT1 isoforms with TGF-β receptors leads to functional crosstalk between two signaling pathways in epithelial ovarian cancer

J Exp Clin Cancer Res371032018

10.1186/s13046-018-0773-8

29751820

Wei

Liu

Zhu

Stat6 cooperates with Sp1 in controlling breast cancer cell proliferation by modulating the expression of p21(Cip1/WAF1) and p27 (Kip1)

Cell Oncol (Dordr)3679932013

10.1007/s13402-012-0115-3

23184467

Morgan

Wasson

Hanson

Kealy

Pentland

McGuire

Scarpini

Coleman

Arthur

JSC

Parish

STAT3 activation by E6 is essential for the differentiation-dependent HPV18 life cycle

PLoS Pathog14e10069752018

10.1371/journal.ppat.1006975

29630659

Hosui

Klover

Tatsumi

Uemura

Nagano

Doki

Mori

Hiramatsu

Kanto

Hennighausen

Suppression of signal transducers and activators of transcription 1 in hepatocellular carcinoma is associated with tumor progression

Int J Cancer131277427842012

10.1002/ijc.27580

22488367

Zhang

Cheng

Liang

Jin

Long non-coding RNA HOTAIR and STAT3 synergistically regulate the cervical cancer cell migration and invasion

Chem Biol Interact2861061102018

10.1016/j.cbi.2018.03.010

29572071

Khan

El-Gamal

A progressive review of V600E-B-RAF-dependent melanoma and drugs inhibiting it

Mini Rev Med Chem173513652017

10.2174/1389557516666160622213142

27337967

Liu

Wang

Huang

Hong

Lin

Tian

Cai

Wang

Leung

Inhibition of the Ras/Raf interaction and repression of renal cancer xenografts in vivo by an enantiomeric iridium(iii) metal-based compound

Chem Sci8475647632017

10.1039/C7SC00311K

28959398

Bosserhoff

Novel biomarkers in malignant melanoma

Clin Chim Acta36728352006

10.1016/j.cca.2005.10.029

16480699

Kugel

IIIAplin

Adaptive resistance to RAF inhibitors in melanoma

Pigment Cell Melanoma Res27103210382014

10.1111/pcmr.12264

24828387

Harpio

Einarsson

S100 proteins as cancer biomarkers with focus on S100B in malignant melanoma

Clin Biochem375125182004

10.1016/j.clinbiochem.2004.05.012

15234232

Xiong

Pan

Expression and clinical significance of S100 family genes in patients with melanoma

Melanoma Res2923292019

10.1097/CMR.0000000000000512

30216200

Figure 1.

KEGG pathway map of the JAK-STAT signaling pathway obtained from the KEGG website. KEGG, Kyoto Encyclopedia of Genes and Genomes; JAK, Janus kinase; STAT, signal transducer and activator of transcription.

Figure 2.

Analysis of enriched GO terms and KEGG pathways for JAK and STAT genes obtained using Database for Annotation, Visualization and Integrated Discovery. (A) Biological process results of GO functional enrichment analysis. (B) Cellular component results of GO functional enrichment analysis. (C) Molecular function results of GO functional enrichment analysis. (D) KEGG pathway analysis results. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; JAK, Janus kinase; STAT, signal transducer and activator of transcription; MAPK, mitogen-activated protein kinase.

Figure 3.

Correlation and association analysis for JAK and STAT genes in SKCM (A) Gene interaction network of JAK and STAT genes generated by GeneMANIA. (B) Protein-protein interaction network generated using the STRING database. (C) Pearson's correlation coefficients for JAK1, STAT1, STAT3, STAT4, STAT5B and STAT6 gene expression levels. (D) Scatter plots of JAK1, STAT1, STAT3, STAT4, STAT5B and STAT6 gene expression level in The Cancer Genome Atlas database. GeneMANIA, gene multiple association network integration algorithm; STRING, Search Tool for the Retrieval of Interacting Genes/Proteins; STAT, signal transducer and activator of transcription; SKCM, skin cutaneous melanoma; error bars in the scatter plots, mean ± SD.

Figure 4.

Prognostic value of the expression of key genes of the JAK-STAT signaling pathway for overall survival. (A-F) Kaplan-Meier survival plots for all patients with skin cutaneous melanoma according to (A) JAK1, (B) STAT1, (C) STAT3, (D) STAT4, (E) STAT5B and (F) STAT6 expression. JAK, Janus kinase; STAT, signal transducer and activator of transcription.

Figure 5.

Nomogram and joint-effects survival analysis of JAK and STAT genes. (A) Nomogram for predicting the 1-, 3- and 5-year event (death) based on risk scores and clinical information. (B) Joint effects analysis of the influence of combined gene expression on overall survival in patients with skin cutaneous carcinoma. (C-G) Joint effects analysis of the influence of combined expression of four selected genes on overall survival in patients with skin cutaneous melanoma according to (C) group (1)-(3), (D) group (4)-(6), (E) group (7)-(9), (F) group (10)-(12) and (G) group (13)-(15). JAK1, Janus kinase 1; STAT, signal transducer and activator of transcription. NOS, non-specific.

Figure 6.

(A-J) Joint effects analysis of the effects of combined expression of three selected genes on overall survival in patients with skin cutaneous melanoma according to (A) group i-iii, (B) group iv-vi, (C) group vi-ix, (D) group x-xii, (E) group xiii-xv, (F) group xvi-xvii, (G) group xix-xxi, (H) group xxii-xxiv, (I) group xxv-xxvii and (J) group xxviii-xxx.

Figure 7.

(A-J) Joint effects analysis of the influence of combined expression of two selected genes on overall survival in patients with skin cutaneous melanoma according to (A) group I–III, (B) group IV–VI, (C) group VI–IX, (D) group X–XII, (E) group XIII–XV, (F) group XVI–XVII, (G) group XIX–XXI, (H) group XXII–XXIV, (I) group XXV–XXVII and (J) group XXVIII–XXX.

Figure 8.

GSEA of genes expressed in patients with skin cutaneous melanoma. (A-D) GSEA of the c5 reference gene sets for the high STAT1 expression group. (A) c5 item negative regulation of adaptive immune response, (B) c5 item positive regulation of cell-cell adhesion, (C) c5 item regulation of CD4 positive alpha beta T cell activation and (D) c5 item regulation of cell-cell adhesion. (E) GSEA of the c5 reference gene sets for the low STAT4 expression group. (F) GSEA of the c5 reference gene sets for the high STAT5B expression group. STAT, signal transducer and activator of transcription; GSEA, gene set enrichment analysis; NES, normalized enrichment score; FDR, false discovery rate; NOM, nominal.

Table I.

Survival analysis based on clinical information.

Characteristic	Patients (n=458)	No. of events (%)	MST (days)	HR (95% CI)	Crude P-value
Race					0.003^a
Caucasian	435	210 (48.3)	2,454	0.337 (0.166–0.687) Ref.
Other	13	8 (61.5)	636
Unknown	10
Sex					0.345
Male	284	146 (51.4)	2,421	Ref. 0.872 (0.657–1.158)
Female	174	73 (42.0)	2,367
Age (years)					<0.001^a
≤60	239	115 (48.1)	3,196	0.587 (0.445–0.773) Ref.
>60	219	104 (47.5)	1,864
Tumor stage^b					<0.001^a
Early	231	108 (46.8)	3,195	0.547 (0.411–0.728) Ref.
Advanced	191	96 (50.3)	1,927
Unknown	36

P<0.05.

Due to the small sample size of certain tumor stages, two groups were used to reduce the probability of statistical errors. MST, median survival time; HR, hazard ratio; CI, confidence interval; Ref., reference group.

Table II.

Prognostic survival analysis based on high or low expression of JAK and STAT family genes

Gene	Patients (n=458)	No. of events (%)	MST (days)	Crude HR (95% CI)	Crude P-value	Adjusted HR^b (95% CI)	Adjusted P-value^b
JAK1
Low	229	99 (43.2)	2,588	Ref. 1.056 (0.808–1.379)	0.690	Ref. 0.950 (0.720–1.253)	0.716
High	229	120 (52.4)	2,365
STAT1
Low	229	119 (52.0)	1,910	Ref 0.587(0.449–0.767)	<0.001^a	Ref. 0.595 (0.455–0.778)	<0.001^a
High	229	100 (47.3)	3,259
STAT3
High	229	113 (49.3)	2,030	Ref. 0.701 (0.537–0.915)	0.009^a	Ref. 0.725 (0.555–0.947)	0.01^8a
Low	229	106 (46.3)	3,080
STAT4
Low	229	121 (52.8)	1,785	Ref. 0.562 (0.430–0.735)	<0.001^a	Ref. 0.590 (0.450–0.773)	<0.001^a
High	229	98 (42.8)	3,176
STAT5B
Low	229	117 (48.1)	2,022	Ref. 0.675 (0.516–0.884)	0.004^a	Ref. 0.690 (0.526–0.940)	0.007^a
High	229	102 (47.4)	3,139
STAT6
High	229	115 (50.4)	2,184	Ref. 0.731 (0.559–0.956)	0.022^a	Ref. 0.737 (0.563–0.964)	0.026^a
Low	229	104 (47.2)	3,139

P<0.05.

Adjusted for age and tumor stage. JAK1, Janus kinase 1; STAT, signal transducer and activator of transcription; MST, median survival time; HR, hazard ratio; CI, confidence interval; Ref., reference.

Table III.

Grouping information for joint effects analysis of all STAT genes.

	Composition

Group	STAT1	STAT3	STAT4	STAT5B	STAT6
1	Low	Low	Low	Low	High
2	High	Low	Low	Low	High
	Low	High	Low	Low	High
	Low	Low	High	Low	High
	Low	Low	Low	High	High
	Low	Low	Low	Low	Low
3	High	High	Low	Low	High
	High	Low	High	Low	High
	High	Low	Low	High	High
	High	Low	Low	Low	Low
	Low	High	High	Low	High
	Low	High	Low	High	High
	Low	High	Low	Low	Low
	Low	Low	High	High	High
	Low	Low	High	Low	Low
	Low	Low	Low	High	Low
4	High	High	High	Low	High
	High	High	Low	High	High
	High	Low	High	High	High
	Low	High	High	High	High
	High	High	Low	Low	Low
	High	Low	High	Low	Low
	Low	High	High	Low	Low
	High	Low	Low	High	Low
	Low	High	Low	High	Low
	Low	Low	High	High	Low
5	High	High	High	High	High
	High	High	High	Low	Low
	High	High	Low	High	Low
	High	Low	High	High	Low
	Low	High	High	High	Low
6	High	High	High	High	Low

Bold, favorable prognosis groups in univariate survival analysis. STAT, signal transducer and activator of transcription.

Table IV.

Grouping information for joint effects analysis of four selected STAT genes.

	Composition

Group	STAT1	STAT3	STAT4	STAT5B	STAT6
(1)	Low	Low	Low	Low	–
(2)	High	Low	Low	Low	–
	Low	High	Low	Low	–
	Low	Low	High	Low	–
	Low	Low	Low	High	–
	High	High	Low	Low	–
	High	Low	High	Low	–
	High	Low	Low	High	–
	Low	High	High	Low	–
	Low	High	Low	High	–
	Low	Low	High	High	–
	High	High	High	Low	–
	High	High	Low	High	–
	High	Low	High	High	–
	Low	High	High	High	–
(3)	High	High	High	High	–
(4)	Low	Low	Low	–	High
(5)	High	Low	Low	–	High
	Low	High	Low	–	High
	Low	Low	High	–	High
	Low	Low	Low	–	Low
	High	High	Low	–	High
	High	Low	High	–	High
	High	Low	Low	–	Low
	Low	High	High	–	High
	Low	High	Low	–	Low
	Low	Low	High	–	Low
	High	High	High	–	High
	High	High	Low	–	Low
	High	Low	High	–	Low
	Low	High	High	–	Low
(6)	High	High	High	–	Low
(7)	Low	Low	–	Low	High
(8)	High	Low	–	Low	High
	Low	High	–	Low	High
	Low	Low	–	High	High
	Low	Low	–	Low	Low
	High	High	–	Low	High
	High	Low	–	High	High
	High	Low	–	Low	Low
	Low	High	–	High	High
	Low	High	–	Low	Low
	Low	Low	–	High	Low
	High	High	–	High	High
	High	High	–	Low	Low
	High	Low	–	High	Low
	Low	High	–	High	Low
(9)	High	High	–	High	Low
(10)	Low	–	Low	Low	High
(11)	High	–	Low	Low	High
	Low	–	High	Low	High
	Low	–	Low	High	High
	Low	–	Low	Low	Low
	High	–	High	Low	High
	High	–	Low	High	High
	High	–	Low	Low	Low
	Low	–	High	High	High
	Low	–	High	Low	Low
	Low	–	Low	High	Low
	High	–	High	High	High
	High	–	High	Low	Low
	High	–	Low	High	Low
	Low	–	High	High	Low
(12)	High	–	High	High	Low
(13)	–	Low	Low	Low	High
(14)	–	High	Low	Low	High
	–	Low	High	Low	High
	–	Low	Low	High	High
	–	Low	Low	Low	Low
	–	High	High	Low	High
	–	High	Low	High	High
	–	High	Low	Low	Low
	–	Low	High	High	High
	–	Low	High	Low	Low
	–	Low	Low	High	Low
	–	High	High	High	High
	–	High	High	Low	Low
	–	High	Low	High	Low
	–	Low	High	High	Low
(15)	–	High	High	High	Low

‘−’, gene not selected. Bold, favorable prognosis groups in univariate survival analysis. STAT, signal transducer and activator of transcription.

Table V.

Grouping information for joint effects analysis of three selected STAT genes.

	Composition

Group	STAT1	STAT3	STAT4	STAT5B	STAT6
i	Low	Low	Low	–	–
ii	High	Low	Low	–	–
	Low	High	Low	–	–
	Low	Low	High	–	–
	High	High	Low	–	–
	High	Low	High	–	–
	Low	High	High	–	–
iii	High	High	High	–	–
iv	Low	Low	–	Low	–
v	Low	Low	–	Low	–
	High	High	–	Low	–
	High	High	–	High	–
	Low	High	–	Low	–
	Low	Low	–	High	–
	High	High	–	High	–
vi	High	High	–	High	–
vii	Low	Low	–	–	High
viii	High	Low	–	–	High
	Low	High	–	–	High
	Low	Low	–	–	Low
	High	High	–	–	High
	High	Low	–	–	Low
	Low	High	–	–	Low
ix	High	High	–	–	Low
x	Low	–	Low	Low	–
xi	High	–	Low	Low	–
	Low	–	High	Low	–
	Low	–	Low	High	–
	High	–	High	Low	–
	High	–	Low	High	–
	Low	–	High	High	–
xii	High	–	High	High	–
xiii	Low	–	Low	–	High
xiv	High	–	Low	–	High
	Low	–	High	–	High
	Low	–	Low	–	Low
	High	–	High	–	High
	High	–	Low	–	Low
	Low	–	High	–	Low
xv	High	–	High	–	Low
xvi	Low	–	–	Low	High
xvii	High	–	–	Low	High
	Low	–	–	High	High
	Low	–	–	Low	Low
	High	–	–	High	High
	High	–	–	Low	Low
	Low	–	–	High	Low
xviii	High	–	–	High	Low
xiv	–	Low	Low	Low	–
xx	–	High	Low	Low	–
	–	Low	High	High	–
	–	Low	Low	Low	–
	–	High	High	High	–
	–	High	Low	Low	–
	–	Low	High	High	–
xxi	–	High	High	High	–
xxii	–	Low	Low	–	High
xxiii	–	High	Low	–	High
	–	Low	High	–	High
	–	Low	Low	–	Low
	–	High	High	–	High
	–	High	Low	–	Low
	–	Low	High	–	Low
xxiv	–	High	High	–	Low
xxv	–	Low	–	Low	High
xxvi	–	High	–	High	High
	–	Low	–	Low	High
	–	Low	–	Low	Low
	–	High	–	High	High
	–	High	–	High	Low
	–	Low	–	Low	Low
xxvii	–	High	–	High	Low
xxviii	–	–	Low	Low	High
xxix	–	–	High	High	High
	–	–	Low	Low	High
	–	–	Low	Low	Low
	–	–	High	High	High
	–	–	High	High	Low
	–	–	Low	Low	Low
xxx	–	–	High	High	Low

‘−’, gene not selected. Bold, favorable prognosis groups in univariate survival analysis. STAT, signal transducer and activator of transcription.

Table VI.

Grouping information for joint effects analysis of two selected STAT genes

	Composition

Group	STAT1	STAT3	STAT4	STAT5B	STAT6
I	Low	Low	–	–	–
II	High	Low	–	–	–
	Low	High	–	–	–
III	High	High	–	–	–
IV	Low	–	Low	–	–
V	High	–	Low	–	–
	Low	–	High	–	–
VI	High	–	High	–	–
VII	Low	–	–	Low	–
VIII	Low	–	–	High	–
	High	–	–	Low	–
IX	High	–	–	High	–
X	Low	–	–	–	High
XI	Low	–	–	–	Low
	High	–	–	–	High
XII	High	–	–	–	Low
XIII	–	Low	Low	–	–
XIV	–	High	Low	–	–
	–	Low	High	–	–
XV	–	High	High	–	–
XVI	–	Low	–	Low	–
XVII	–	Low	–	High	–
	–	High	–	Low	–
XVIII	–	High	–	High	–
XIX	–	Low	–	–	High
XX	–	Low	–	–	Low
	–	High	–	–	High
XXI	–	High	–	–	Low
XXII	–	–	Low	Low	–
XXIII	–	–	Low	High	–
	–	–	High	Low	–
XXIV	–	–	High	High	–
XXV	–	–	Low	–	High
XXVI	–	–	Low	–	Low
	–	–	High	–	High
XXVII	–	–	High	–	Low
XXVIII	–	–	–	Low	High
XXIV	–	–	–	Low	Low
	–	–	–	High	High
XXX	–	–	–	High	Low

‘−’, gene not selected. Bold, favorable prognosis groups in univariate survival analysis. STAT, signal transducer and activator of transcription.

Table VII.

Joint effects analysis of the prognostic value of combinations of gene expression in skin cutaneous melanoma.

Group	No. of genes	Patients (n=458)	No. of events (%)	MST (days)	Log-rank P-value	HR (95% CI)
1	5	38	24 (63.1)	1,301	Ref.	Ref.
2	5	87	46 (52.9)	2,030	0.073	0.636 (0.487–1.043)
3	5	106	53 (50.0)	1,766	0.094	0.661 (0.407–1.073)
4	5	98	37 (37.8)	3,196	<0.001^a	0.352 (0.210–0.590)
5	5	93	42 (45.2)	4,533	<0.001^a	0.322 (0.194–0.535)
6	5	36	17 (47.2)	3,259	<0.001^a	0.280 (0.149–0.528)
(1)	4	71	42 (59.2)	1,441	Ref.	Ref.
(2)	4	301	140 (46.4)	2,192	0.014^a	0.648 (0.458–0.915)
(3)	4	86	37 (43.0)	4,930	<0.001^a	0.376 (0.241–0588)
(4)	4	61	37 (60.7)	1,197	Ref.	Ref.
(5)	4	351	160 (45.6)	2,829	<0.001^a	0.515 (0.360–0.738)
(6)	4	46	22 (47.8)	5,237	<0.001^a	0.312 (0.182–0.535)
(7)	4	47	27 (57.4)	1,354	Ref.	Ref.
(8)	4	365	171 (46.8)	2,470	0.002^a	0.530 (0.352–0.798)
(9)	4	46	21 (45.7)	3,266	<0.001^a	0.370 (0.189–0.601)
(10)	4	56	34 (60.7)	1,354	Ref.	Ref.
(11)	4	351	160 (45.6)	2,588	<0.001^a	0.484 (0.333–0.704)
(12)	4	51	25 (49.0)	3,259	<0.001^a	0.338 (0.200–0.573)
(13)	4	42	26 (61.9)	1,354	Ref.	Ref.
(14)	4	372	171 (46.0)	2,588	0.001^a	0.498 (0.328–0.755)
(15)	4	44	22 (50.0)	2,927	<0.001^a	0.344 (0.193–0.612)
i	3	130	75 (57.7)	1,441	Ref.	Ref.
ii	3	250	112 (44.8)	2,273	0.022^a	0.710 (0.529–0.952)
iii	3	103	46 (44.7)	4,930	<0.001^a	0.418 (0.289–0.605)
iv	3	91	47 (51.6)	1,949	Ref.	Ref.
v	3	267	131 (49.1)	2,071	0.086	0.746 (0.534–1.042)
vi	3	125	51 (40.8)	4,930	<0.001^a	0.415 (0.278–0.620)
vii	3	78	44 (56.4)	1,197	Ref.	Ref.
viii	3	312	146 (46.8)	2,470	<0.001^a	0.530 (0.377–0.744)
ix	3	93	39 (41.9)	4,526	<0.001^a	0.328 (0.212–0.508)
x	3	103	61 (59.2)	1,487	Ref.	Ref.
xi	3	245	127 (51.8)	1,864	0.127	0.788 (0.580–1.070)
xii	3	135	60 (44.4)	4,533	<0.001^a	0.388 (0.270–0.557)
xiii	3	95	55 (57.9)	1,354	Ref.	Ref.
xiv	3	276	152 (551)	2,889	<0.001^a	0.480 (0.348–0.661)
xv	3	112	50 (44.6)	3,259	<0.001^a	0.367 (0.249–0.542)
xvi	3	69	40 (58.0)	1,429	Ref.	Ref.
xvii	3	324	148 (45.7)	2,588	<0.001^a	0.502 (0.352–0.715)
xviii	3	90	40 (44.4)	3,266	<0.001^a	0.358 (0.229–0.559)
xix	3	87	51 (58.6)	1,655	Ref.	Ref.
xx	3	268	121 (45.1)	2,367	0.012^a	0.658 (0.474–0.913)
xxi	3	128	57 (44.5)	4,507	<0.001^a	0.419 (0.286–0.613)
xxii	3	70	43 (61.4)	1,301	Ref.	Ref.
xxiii	3	328	148 (45.1)	2,948	<0.001^a	0.530 (0.377–0.745)
xxiv	3	85	38 (44.7)	3,259	<0.001^a	0.348 (0.223–0.542)
xxv	3	66	35 (53.0)	1,910	Ref.	Ref.
xxvi	3	323	150 (46.4)	2,470	0.043^a	0.683 (0.472–0.988)
xxvii	3	94	44 (46.8)	3,266	0.001^a	0.462 (0.294–0.724)
xxviii	3	70	40 (57.1)	1,486	Ref.	Ref.
xxix	3	323	149 (46.1)	2,470	0.001^a	0.536 (0.376–0.764)
xxx	3	90	40 (44.4)	5,237	<0.001^a	0.349 (0.223–0.545)
I	2	139	70 (50.4)	1,949	Ref.	Ref.
II	2	180	92 (51.1)	2,005	0.519	0.902 (0.660–1.233)
III	2	164	67 (40.9)	4,526	<0.001^a	0.485 (0.346–0.678)
IV	2	168	96 (57.1)	1,487	Ref.	Ref.
V	2	122	48 (39.3)	2,454	0.017^a	0.655 (0.462–0.928)
VI	2	193	85 (44.0)	3,564	<0.001^a	0.477 (0.356–0.641)
VII	2	133	70 (52.6)	1,780	Ref.	Ref.
VIII	2	192	90 (46.9)	2,273	0.008^a	0.654 (0.477–0.897)
IX	2	133	59 (44.4)	4,507	<0.001^a	0.449 (0.316–0.638)
X	2	125	70 (56.0)	1,438	Ref.	Ref.
XI	2	208	94 (45.2)	3,080	<0.001^a	0.483 (0.353–0.662)
XII	2	125	55 (44.0)	3,259	<0.001^a	0.435 (0.304–0.623)
XIII	2	130	75 (57.7)	1,441	Ref.	Ref.
XIV	2	198	84 (42.4)	2,545	0.011^a	0.677 (0.488–0.912)
XV	2	130	60 (46.2)	3,259	<0.001^a	0.466 (0.332–0.656)
XVI	2	148	72 (48.6)	1,992	Ref.	Ref.
XVII	2	162	80 (49.4)	2,071	0.297	0.843 (0.612–1.162)
XVIII	2	148	67 (45.3)	3,259	0.001^a	0.564 (0.403–0.791)
XIX	2	111	61 (54.9)	1,910	Ref.	Ref.
XX	2	236	106 (44.9)	3,080	0.008^a	0.652 (0.476–0.894)
XXI	2	111	52 (46.8)	3,259	<0.001^a	0.508 (0.349–0.739)
XXII	2	141	77 (54.6)	1,780	Ref.	Ref.
XXIII	2	176	78 (44.3)	2,192	0.015^a	0.674 (0.490–0.926)
XXIV	2	141	64 (45.4)	3,259	<0.001^a	0.466 (0.333–0.653)
XXV	2	118	65 (55.1)	1,544	Ref.	Ref.
XXVI	2	222	106 (47.4)	3,136	<0.001^a	0.548 (0.400–0.750)
XXVII	2	118	48 (40.7)	3,259	<0.001^a	0.399 (0.273–0.583)
XXVIII	2	105	56 (53.3)	1,780	Ref.	Ref.
XXIX	2	248	114 (46.0)	2,470	0.013^a	0.667 (0.484–0.920)
XXX	2	105	49 (46.7)	3,424	<0.001^a	0.466 (0.314–0.692)

P<0.05. MST, median survival time; HR, hazard ratio; CI, confidence interval; Ref., reference.