Identification of URGs clusters in ccRCCAs shown in Fig. 1A and B, a consensus clustering method was used to analyze the correlation between URG expression and ccRCC subtypes. Based on the CDF curve, two clusters (C1 and C2) were formed. The C2 individuals had remarkably poor prognosis compared with C1 (Fig. 1C). At the same time, we found that patients in the high-risk group had lower progression free survival (PFS) (Supplementary Figure S1). Figure 1D illustrates the connection among URGs clusters, URGs expression, and clinical features in ccRCC patients.Fig. 1URGs clusters and clinical characteristics between ccRCC samples in two clusters. (A) The cumulative distribution function curve illustrates the most effective way of URGs clustering. (B) The consensus matrix of the clustering analysis via k-means clustering (k = 2). (C) Kaplan–Meier (KM) curves for the overall survival (OS) of ccRCC patients among different URGs groups. (D) Heatmap of URGs expression in ccRCC patients with different clinical characteristics and URGs clusters. (E) The differences in immune cell infiltration between two clusters.The assessment of immune cell levels in two clusters was conducted due to the correlation between the variation of immune cells and the onset and progression of ccRCC. Cluster 1 exhibited elevated levels of myeloid dendritic cells, neutrophils, and T cells compared to cluster 2, whereas fibroblasts, CD8 T cells, and cytotoxic lymphocytes were lower in C1 than in C2 (Fig. 1E).Development of URGs signature in ccRCCBy using the “limma” package, DEGs between two clusters were identified based on FDR <0.05 and |log fold change (FC)| >1. Next, the univariate Cox analysis was performed to determine prognosis-related DEGs. Afterwards, we conducted a LASSO examination and multiCox to remove the genes that were excessively fitted, resulting in the formation of a signature of six genes (PDK4, PLAUR, UCN, RNASE2, KISS1, and MXD3) known as the URGs (Fig. 2A and B). The equation applied to derive the risk score is as indicated: risk score= (PDK4 × (-0.171141153581758) + (PLAUR × (0.22368544341121) + (UCN × (0.50294554014386) + (RNASE2 × (0.300559387539641) + (KISS1 × (0.312742555297056) + (MXD3 × (0.427843296818388).Fig. 2Construction of the prognostic signature. (A) LASSO coefficient profiles (y-axis) of the gene sets and the optimal penalization coefficient (l) via 3-fold cross-validation based on partial likelihood deviance. (B) The dotted vertical lines represent the optimal values of l. The top x-axis has the numbers of gene sets, whereas the lower x-axis revealed the log (λ). (C, D) Risk score and survival outcome of each case. (E) Heatmap showed the expression of risk genes in two risk groups. (F) PCA of high- and low-risk groups. (G) The KM curve showed that patients in the high-risk group had a worse prognosis. (H) The AUC for 1-, 3- and 5-years survival.Patients diagnosed with ccRCC were divided into low- and high-risk categories based on the median of risk score (Fig. 2C and D). Figure 2E demonstrates the different manifestations of these six genes in subcategories. The results of principal component analysis (PCA) indicated that patients could be effectively categorized into groups with high and low-risk, as demonstrated in Fig. 2F. Moreover, ccRCC with elevated risk scores exhibited a higher mortality rate (Fig. 2G). Furthermore, the ROC curve indicated that the AUC values for the 1-year, 3-year, and 5-year durations were 0.845, 0.779, and 0.792 correspondingly, as shown in Fig. 2H.Validation of URGs prognostic signatureThen the URGs signature was verified in test datasets. Figure 3A-C classified all patients in the test datasets into low- and high-risk groups. In TCGA-test, TCGA-all, and E-MTAB-1980, the K-M analysis revealed that patients with low-risk had a more positive outlook compared to those with high risk (Fig. 3D-F). In the TCGA-test cohort (Fig. 3G), the AUC values for the 1-, 3-, and 5-year periods were 0.777, 0.733, and 0.770, respectively. Similarly, in the TCGA-all cohort (Fig. 3H), the AUC values were 0.798, 0.756, and 0.7737. Additionally, the E-MTAB-1980 cohort (Fig. 3I) had AUC values of 0.760, 0.797, and 0.761. Figure 3J-L demonstrates the different manifestations of risk genes in subcategories. In addition, survival analysis was performed on the clinical features, indicating that individuals with lower risk scores had significantly better outcomes compared to high-risk groups in various categories including age < = 65, T1-2, T3-4, age > 65, female, male, G1-2, G3-4, M0, M1, Stage I + II, Stage III-IV, and N0 (Fig. 4A-B). This further confirmed the validity of the risk model we developed. Furthermore, we conducted a comparison with the prognostic model of different individuals, and it was observed that the C-index of URGs signature surpassed that of others (Fig. 4C). The expression of these six genes was verified using RT-qPCR. Higher expression of PLAUR, UCN, RNASE2, KISS1 and MXD3 in tumor than normal tissue was found, while the expression of PDK4 was lower in tumor tissue (Supplementary Figure S2).Fig. 3Validation of the prognostic signature. Risk score and survival outcome of each case in TCGA-test (A), TCGA-all (B), and E-MTAB-1980 (C). KM curve showed that patients in the high-risk group had a worse prognosis in TCGA-test (D), TCGA-all (E), and E-MTAB-1980 (F). The AUC for 1-, 3- and 5-years survival in TCGA-test (G), TCGA-all (H), and E-MTAB-1980 (I). Heatmap showed the expression of risk genes in TCGA-test (J), TCGA-all (K), and E-MTAB-1980 (L).Fig. 4Survival curve after grouping according to clinical features, including age < = 65, > 65, Female, Male, G1-2, G3-4, M0, M1, N0, StageI-II, StageIII-IV, T1-2, T3-4 (A-B). (C) C-index of our signature and others.Analysis of URGs signature with clinical characteristics and mutationThe association between the risk score and T stage, N stage, M stage, histological grade, and tumor stage was discovered. The risk score gradually increased as the T stage, N stage, M stage, pathological stage, and histological grade deteriorated. Nevertheless, gender and age (Fig. 5A-G) did not show any substantial correlation with risk scores. Figure 5H-I displayed the list of the top 20 genes exhibiting mutations within subgroups. Linear regression analysis (Fig. 5J) revealed that the high-risk group exhibited higher TMB scores compared to the low-risk group, indicating a strong positive correlation between TMB and risk score (Fig. 5K). According to the analysis of the KM curve, the combination of TMB and risk score indicated that individuals with a high risk and high TMB score experienced an unfavorable prognosis (Fig. 5L).Fig. 5URGs signature correlates with clinical features. (A-G) Risk scores among different clinical features of ccRCC, including histological grade, tumor stage, sex, age, T stage, N stage and M stage in TCGA cohort. (H-I) The waterfall plot of tumor somatic mutation displayed distribution of top 20 highly mutated genes in the high-risk and low-risk groups. (J) Scatter plots depicting the correlation between risk score and TMB. (K) Box plot showing the correlation of TMB with risk score. (L) The KM curve of combination of TMB and risk score.Construction of a nomogram for ccRCCIn both Multivariate and univariate Cox regression analyses (Fig. 6A and B), it was demonstrated that the risk score served as a separate prognostic indicator (P < 0.001). The innovative nomogram, developed with URGs signature and clinical parameters, exploited prognostic potential of the URGs signature (Fig. 6C). Next, we depicted the corresponding calibration graphs for 1, 3, and 5 years, and the calibration line exhibited outstanding performance (Fig. 6D). Furthermore, the nomogram’s ability to predict prognosis was evaluated through ROC analysis. Figure 6E shows that the AUC value for 1-year survival periods was 0.913 (nomogram) and 0.797 (risk score). The AUC value for 3-year survival periods was 0.845 (nomogram) and 0.787 (risk score) (Fig. 6F). The AUC value for 5-year survival periods was 0.858 (nomogram) and 0.831 (risk score) (Fig. 6G). At the same time, the calibration plots and ROC showed that the model also had good predictive efficacy in the validation sets (Supplementary Figure S3A-D). Our findings showed the immense potential of this innovative nomogram to serve as an outstanding prognosis prediction model.Fig. 6Construction and assessment of nomogram. (A) Univariate Cox regression (B) multivariate Cox regression analyses. (C) The prediction of nomogram in the TCGA dataset. (D) Calibration plots for the nomogram. The multifactor AUC for 1- (E), 3- (F), and 5-years (G) survival.Analyzing the TME in high- and low-risk groupsImmune cells have a significant role in the tumor microenvironment (TME), which is a key indication of tumor biological behavior. The differences in levels of immune cells between the two risk groups were examined using CIBERSORT, MCPCOUNTER, QUANTISEQ, EPIC, TIMER, CIBERSORT-ABS, and XCELL. In the low-risk category, the levels of neutrophil cells, macrophage, CD4 T cells, NK cells, and B cells were individually elevated, as depicted in Fig. 7A. The examination of immune checkpoint-related genes’ expression patterns was conducted in patients categorized as low- and high-risk. The high-risk group exhibited higher expression levels of the majority of genes related to immune checkpoints (Fig. 7B). In the ssGSEA analysis, the low-risk patients showed reduced infiltration of aDCs, CD8 + T cells, T helper cells, Tumor-infiltrating cell (TIL), and T cells regulatory (Treg), whereas the low-risk group exhibited elevated levels of mast cells and iDCs (Fig. 7C). Moreover, the high-risk patients exhibited enhancements in various immunologic functions, such as T cell co-stimulation, CCR, Type I IFN response, and T cell co-inhibition, as depicted in Fig. 7D. The TIDE score is a crucial metric for forecasting the impact of immunotherapy. According to Fig. 7E, the low-risk category exhibits a greater TIDE score compared to the high-risk category, indicating that patients at high risk demonstrate a more favorable reaction to immunotherapy. Furthermore, Fig. 7F illustrated the spread of individuals with low and high-risk among various immune subtypes.Fig. 7Analysis of immune conditions of high- and low-risk groups. (A) The analysis of differences in immune cell infiltration between the two groups with multiple algorithms. (B) The immune checkpoint-related genes expression levels in different groups. (C) The analysis of differences in immune cell infiltration between the two groups with ssGSEA. (D) The analysis of differences in immune functions between the two groups with ssGSEA. (E) The TIDE scores in different groups. (F) The distribution of patients with high- and low-risk in different immune subtypes.Functional enrichment of the URGs signatureThe GO and KEGG enrichment analyses were performed to explore the underlying biological functions of the URGs signature. The GO analysis revealed that the genes showing differential expression between low- and high-risk groups were mainly enriched in humoral immune response, immunoglobulin complex, and binding to immunoglobulin receptors (Fig. 8A). In the meantime, the KEGG analysis showed that the genes with differential expression were mainly concentrated in the interaction between cytokines and cytokine receptors, the IL-17 signaling pathway, and the cascades of complement and coagulation (Fig. 8B). The DO result showed these DEGs were mainly enriched in mouth disease, urinary system disease and kidney disease (Fig. 8C). Furthermore, there was a significant alteration observed in numerous pathways when comparing the high-risk and low-risk categories, as indicated by the GSVA results (Fig. 8D).Fig. 8Function analysis. (A) GO analysis of DEGs between high and low-risk groups. (B) KEGG analysis of DEGs between high- and low-risk groups. (C) DO analysis of DEGs. (D) GSVA enrichment analysis in high- and low-risk groups.Analyzing the correlation between drug sensitivity and risk scoreTo evaluate potential differences in drug sensitivity between low- and high-risk groups, we examined the correlation between the risk scores of ccRCC patients and the IC50 values of chemotherapy and targeted treatment drugs. Patients classified as low-risk exhibited greater sensitivity to four drugs (AKT inhibitor VIII, AS601245, BX-912, and QS11) compared to the high-risk group, as evidenced by their lower IC50 values. In Fig. 9, the high-risk group exhibited lower IC50 values for five drugs (CGP-60474, CP466722, CNF-2, WZ3105, and YM155) compared to the low-risk group.Fig. 9Drug sensitivity analysis in high and low-risk groups.
