RNA-seq data and clinical parameters of different cancers and matched normal tissues were collected from TCGA database. The RNA-seq data were then normalized via the UCSC Xena tool (https://xena.ucsc.edu/). Genetic profiles of 31 human normal tissues were obtained from GTEx (https://commonfund.nih.gov/GTEx). Duplicated data were excluded.

The prognostic value of ETV4 in various types of cancers was analyzed using univariate Cox analysis and Kaplan-Meier survival analysis through the online tool Sangerbox (http://vip.sangerbox.com/). Data for expression analysis were from TCGA and GTEx database. Each expression value was transformed via log2(x+1) and single cancer with less than three samples was removed. Unpaired Student’s t-Tests were used for differential significance analysis. The clinical parameters used for survival analysis were obtained from TCGA database. Cox regression and Kaplan-Meier survival analysis were used for survival outcome analysis.

Immunohistochemistry images of ETV4 in tumors and non-tumor specimens were acquired from the Human Protein Atlas (http://www.proteinatlas.org/).

Correlations between ETV4 expression levels and immune checkpoints, ImmuneScore, StromalScore, and ESTIMATEScore were calculated using the online tool Sangerbox with Pearson’s analysis. Data used for the analysis were downloaded from TCGA database.The relationship between ETV4 levels and immune-infiltrating cells (CD4+ T cells, CD8+ T cells, B cells, neutrophils, and dendritic cells) was calculated using the online tool Tumor Immune Estimation Resource (TIMER, https://cistrome.shinyapps.io/timer/) with Pearson’s analysis. A p-value <0.05 was regarded as statistically significant.

The relationships between ETV4 expression and tumor heterogeneity and stemness were assessed among all TCGA tumor specimens using the online tool Sangerbox with Pearson’s correlation analysis. mRNA and DNA methylation profiles were used to calculate the six stemness scores (ss): RNA expression-based (RNAss), epigenetically regulated RNA expression-based (EREG.EXPss), DNA methylation-based (DNAss), epigenetically regulated DNA methylation-based (EREGMETHss), differentially methylated probes-based (DMPss), and enhancer elements/DNA methylation-based (ENHss) according to previous research (17). These stemness indices range from 0 to 1, where 0 means that the tumor cells are less similar to stem cells, and 1 means that the tumor cells are more similar to stem cells.

Expression data for five MMR genes (PMS2, MLH1, EPCAM, MSH2, and MSH6) were obtained from TCGA database and correlated with ETV4 expression using Spearman’s correlation. A similar method was also employed to assess the correlation between ETV4 levels and four DNA methyltransferases (DNMT1, DNMT2, DNMT3A, DNMT3B).

To evaluate the impact of ETV4 levels on drug sensitivity, drug sensitivity and gene expression data were downloaded from the CellMiner database (https://discover.nci.nih.gov/cellminer/) which is a large database for collecting, processing, and integrating molecular data on NCI-60 and other tumor cells. Cell sensitivity to a drug was indicated by the z-score in the compound activity profile; the drug’s anticancer activity increased as the z-score value increased. Data were analyzed using Pearson’s correlation using R (4.2.1) software. Drugs were only included if they were currently in a clinical trial or had been approved by the FDA. A p-value <0.01 was regarded as statistically significant. Findings were visualized using “ggplot2” and “ggpubr” packages.

To investigate the differential expression of ETV4 between normal tissues and tumor specimens, mRNA levels were first compared between tissue types using RNA-seq data from TCGA database. As shown in Figure 1A and Supplementary Table 1 , in comparison with normal tissues, ETV4 expression was remarkably elevated in 17 tumors such as glioblastoma multiforme (GBM), lung adenocarcinoma (LUAD), colon adenocarcinoma (COAD), esophageal carcinoma, and colon adenocarcinoma/rectum adenocarcinoma esophageal carcinoma (COADREAD). In some tumors, ETV4 expression was lower than in normal tissues: kidney renal papillary cell carcinoma (KIRP), prostate adenocarcinoma (PRAD), pan-kidney cohort (KIPAN), kidney renal clear cell carcinoma (KIRC), and pheochromocytoma and paraganglioma (PCPG). Because TCGA dataset comprises mainly tumor tissue and very little normal tissue, ETV4 expression was reanalyzed using data from both TCGA and GTEx. As shown in Figure 1B , the consolidated result showed that ETV4 was markedly elevated in 28 tumors (including GBM and LUAD) and still downregulated in KIRP, KIPAN, PRAD, KIRC, and PCPG. Similar results were also observed for the immunohistochemical tissue assay ( Figure 1C ). These data reveal that over-expression of ETV4 in most cancer tissues compared to normal tissues.

To better understand the prognostic value of ETV4 in different cancer types, we investigated correlations between ETV4 expression and survival outcomes for each cancer using Cox regression and Kaplan-Meier survival analysis. As seen in Figures 2A , 3A , patients suffering glioma (GBMLGG), brain low grade glioma (LGG), sarcoma, kidney renal papillary cell carcinoma, head and neck squamous cell carcinoma, GBM, KIRC, mesothelioma (MESO), liver hepatocellular carcinoma (LIHC), and adrenocortical carcinoma (ACC) with high expression of ETV4 had worse overall survival. Association analysis of disease-free-interval showed that ETV4 expression was a hazard factor in lung squamous cell carcinoma, COADREAD, COAD, and PCPG, but was a favorable factor in esophageal carcinoma ( Figure 2B and Figure 3B ). High expression of ETV4 predicted poorer progression-free interval for GBMLGG, ACC, MESO, PCPG, LGG, and cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), and better progression-free interval for ovarian serous cystadenocarcinoma ( Figures 2C , 3C ). High expression of ETV4 was remarkably correlated with poor disease-specific survival in GBMLGG, ACC, LGG, KIRC, GBM, LIHC, KIRP, MESO, and sarcoma, and with improved prognosis in thyroid carcinoma (THCA) ( Figures 2D , 3D ). The above data indicate that ETV4 may take a pivotal role in the survival of patients with GBMLGG, ACC, LGG, KIRC, GBM, LIHC, KIRP, MESO, sarcoma and THCA.

The tumor microenvironment plays an important role in tumor initiation, metastasis, progression, and therapeutic outcomes (18, 19) and it has been confirmed that balance of immune checkpoints and immune cells in the TME is critical to suppress tumor development and progression (20, 21). To investigate the function of ETV4 in the tumor microenvironment, we calculated the correlation between ETV4 levels and 60 common immune checkpoint (ICP) genes (36 stimulatory and 24 inhibitory). As shown in Figure 4A , the relationship between ETV4 and ICP genes varied among different cancer types. ETV4 expression was positively associated with majority of ICP genes in many cancers, such as ovarian cancer and PCPG, but negatively correlated in KIPAN and COAD.This finding implied that ETV4 might promote tumorigenesis and progression via breaking the balance of ICP. Next, relationship between ETV4 levels and immune cell infiltration (comprising B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells) was also investigated. As shown in Figure 4B , ETV4 expression was significantly negative correlated with most of immune cell infiltration in cancers. Additionally, ImmuneScore and StromalScore were also employed to evaluate the relationship between ETV4 expression and immune cell infiltration among cancers. As shown in Supplementary Figure S1 , ETV4 expression was positively related to StromalScore in GBMLGG, PRAD, thymoma, and PCPG, and negatively related to StromalScore in 19 other tumors. ETV4 expression was associated positively with ImmuneScore and ESTIMATEScore in PRAD and PCPG, and negatively in GBM, LGG, and CESC. The three tumors most strongly correlated with ETV4 expression were PRAD, COAD, stomach adenocarcinoma (STAD) (StromalScore) ( Figure 5A ); PRAD,COAD, STAD (ImmuneScore) ( Figure 5B ); and PRAD, COAD, STAD (ESTIMATEScore) ( Figure 5C ). These results reveal that ETV4 may promote cancer progression via regulating immune checkpoint genes expression and immune cells infiltration.

Heterogeneity is a distinct feature of malignancies. After several divisions and proliferation, their descendants acquire genes or altered molecular biology that lead to further differences in proliferation rates, capacity for invasion and metastasis, or drug resistance. Tumor heterogeneity can be assessed by microsatellite instability (MSI), tumor mutational burden (TMB), homologous recombination deficiency (HRD), tumor purity, neoantigen (NEO), loss of heterozygosity (LOH), tumor ploidy, and mutant-allele tumor heterogeneity (MATH). RNA-seq data from TCGA were used to assess the relationship between ETV4 expression and cancer heterogeneity. As shown in Figure 6 , ETV4 was positively associated with TMB in MESO and ACC and negatively related to TMB in LUAD, COAD, COADREAD, KIRP, and CHOL. ETV4 was positively associated with MATH in 14 tumors and negatively associated with MATH in GBMLGG, LGG, and skin cutaneous melanoma. ETV4 was positively associated with MSI in CESC, KIPAN, MESO, skin cutaneous melanoma and kidney chromophobe and negatively associated with MSI in COAD, COADREAD, stomach and esophageal carcinoma, STAD and GBMLGG. ETV4 was negatively associated with NEO in LUAD, COAD, COADREAD, testicular germ cell tumors, and CHOL. It was positively associated with tumor purity in 19 tumors and negatively associated with purity in PRAD, LIHC, testicular germ cell tumors, and PCPG. ETV4 was positively associated with ploidy in nine tumors and negatively associated with ploidy in GBMLGG, LGG, KIPAN, THCA, and ovarian serous cystadenocarcinoma. It was related positively to HRD in 14 tumors and negatively in GBMLGG and LGG. ETV4 was positively correlated with LOH in 18 tumors and negatively correlated with LOH in LUAD and THCA. Together these results indicate that ETV4 may accelerate tumor genesis and prognosis by regulating tumor heterogeneity.

The acquisition of stem cell-like features is a crucial driver of tumor progression (22). Stemness indices rely on DNA methylation and mRNA expression are calculated to reflect the resemblance between cancer cells and stem cells (17). The indices include RNAss, EREG EXPss, DNAss, EREG METHss, DMPss, and ENHss, each of which varies from zero (least resemblance to stem cells) to one (most resemblance to stem cells). ETV4 expression was compared to stemness indices across various tumor types. As shown in Figure 7 , ETV4 levels were significantly correlate with RNAss index in 18 cancers (positively associated in 14cancers), EREG EXPss in 20cancers (positively associated in 14cancers), DNAss in 17 cancers (positively associated in16 cancers), EREG METHss in 17 cancers (positively associated in 16 cancers), DMPss in 16 cancers (positively associated in 13 cancers), and ENHss in 15 cancers (positively associated in 14cancers) respectively. These results suggest that ETV4 may be related to the genesis of tumor stemness in most cancers.

Disorder of MMR genes and DNA methyltransferases plays a pivotal role in tumorigenesis and prognosis (23, 24). To investigate the role of ETV4 in tumorigenesis, ETV4 was correlated with five different MMR genes. As shown in Figure 8A , ETV4 was significantly associated with MMR genes in 26 tumors (notably, it was not associated with MMR genes in acute myeloid leukemia, ovarian serous cystadenocarcinoma, PRAD, READ, thymoma, uterine carcinosarcoma, and uveal melanoma). ETV4 was then correlated with four DNA methyltransferases. As shown in Figure 8B , ETV4 expression was highly associated with these four methyltransferases in many cancers. These results suggest that ETV4 may promote tumorigenesis and poor prognosis by modulating DNA repair genes and DNA methyltransferases.

The relationship between ETV4 levels and drug sensitivity was assessed using data from CellMiner. The 16 drugs with the strongest correlations are shown in Figure 9 . Notably, ETV4 levels were negatively correlated with drug sensitivity of GSK-60903, AZD-2858, Futibatinib, ON-123300, LY-2090314 and positively associated with sensitivity of BGB-283, TAK-632, MLN-2480, AZ-628, LXH-254, LY-3009120, Dabrafenib, Refametinib, Selumetinib, PLX-4720, and PLX-8394. These findings suggest that ETV4 levels can be used for anticancer drug selection.