Expression levels of CDCA7 in ccRCC
We analyzed the mRNA expression levels of CDCA7 in 531 ccRCC tissues and 72 normal kidney tissues from the TCGA dataset and found that CDCA7 was upregulated in ccRCC tissues compared to the normal tissues (P<0.001, Figure 1A-1B). The pairwise boxplot of 72 pairs of ccRCC tissues and matched adjacent normal tissues from TCGA showed most of the cancer tissues exhibited a higher level of CDCA7 (P<0.001, Figure 1C). We further verified the high expression level of CDCA7 in the International Cancer Genome Consortium (ICGC) dataset as external verification (P<0.001, Figure 1D). Furthermore, qRT-PCR results from 16 pairs of ccRCC and adjacent normal kidney tissues also exhibited a higher expression of CDCA7 in ccRCC tissues (Figure 1E). According to the CDCA7 expression levels of the 531 ccRCC patients, we set the median expression level as the cut-off value and divided these patients into a high- and low-risk group respectively. The Kaplan–Meier curve was plotted and showed that patients in the high-risk group had significantly poorer overall survival than those in the low-risk group (P<0.001, Figure 1F), suggesting its potential to predict ccRCC patients’ prognosis. Moreover, we plotted the Kaplan–Meier curve of CDCA7 in ccRCC patients with the help of ArrayExpress database (E-MTAB-1980) as external verification, which exhibited the same results (P<0.05, Figure 1G). The ROC analysis was performed and its AUC for CDCA7 was 0.661 (Figure 1H), indicating the barely satisfactory prognosis predicted ability of CDCA7.
Association of CDCA7 expression with clinicopathologic parameters
To investigate the association between CDCA7 expression and related clinicopathologic parameters, we analyzed the CDCA7 expression levels in different groups of clinicopathologic characteristics by means of independent sample t-tests. The results showed that the CDCA7 exhibited higher expression levels in groups of higher grade (P<0.001; Figure 2A), pathologic stage (P<0.001; Figure 2B), T stage (P<0.001; Figure 2C), and M stage (P<0.01; Figure 2D). No significantly difference was observed of CDCA7 expression levels in age, gender, ethnicity, and N stage (date not shown).
Univariate and multivariate analysis of OS and construction of ccRCC prognostic prediction nomogram
We carried out univariate Cox and multivariate Cox regression analysis to investigate whether the CDCA7 expression was an independent prognostic factor correlated with the overall survival (OS) of ccRCC patients (Table 1). As showed in Figure 3A, in the univariate Cox regression analysis, CDCA7 expression, grade, age, pathological stage, T stage and M stage were all significantly associated with OS of ccRCC patients. However, in the multivariate Cox regression analysis, only CDCA7 expression, grade, pathological stage, and N stage showed significant correlation with OS of ccRCC patients (Figure 3B), and high CDCA7 expression predicted a poorer OS (HR= 1.125; P<0.001). Based on the results above, CDCA7 could act as an independent prognostic factor of OS when adjusted by other related variables.
Then we carried out ROC curve analysis to assess the predictive ability of CDCA7 and other clinicopathologic parameters (Figure 3C). The AUC of the CDCA7 expression was 0.660, higher than that of age, gender, race and lymph nodes status, and lower than that of tumor grade, pathological stage, T stage and M stage. According to the AUC results, we realized that CDCA7 expression level alone could not sufficiently predict prognosis of ccRCC patients. We further established a prognostic nomogram by integrating CDCA7 and clinicopathologic parameters (Figure 3D). The nomogram could help to evaluate 1-, 3-, and 5-year survival probabilities to predict ccRCC patients’ prognosis with a quantitative approach.
GSEA analysis of CDCA7
Though we discovered CDCA7 could act as an independent prognostic factor in ccRCC, how CDCA7 is involved in the ccRCC pathogenesis still remained unclear. We performed GSEA analysis to explore possible mechanisms and signaling pathways through which CDCA7 functioned to regulate ccRCC. On the basis of the normalized enrichment score (NES) and FDR q-val (FDR<0.01), the most significantly enriched biological pathways were exhibited (Table 2, Figure 4), which were apoptosis pathway, cell cycle pathway, JAK-STAT pathway, NOD like receptor pathway, P53 pathway, T cell receptor pathway and toll like receptor pathway (Figure 4A~H), to uncover the potential regulatory mechanism of CDCA7 in ccRCC.
PPI network construction and association of CDCA7 with MSI, TMB, Neoantigen in ccRCC
To explore the potential functional interaction of CDCA7, we constructed the PPI network by applying the STRING database and the Cytoscape software. As showed in Figure 5A, ten genes including SLBP, GINS2, HELLS, UHRF1, MCM2, MCM4, MCM5, NASP, TYMS, and CDC6 were significantly associated with CDCA7 functionally. Based on the ccRCC samples from TCGA database, we further investigated whether CDCA7 was relevant to MSI, TMB or neoantigen. The results suggested that CDCA7 was significantly related to MSI (P<0.001, Figure 5B) and TMB (P<0.001, Figure 5C), while it was not associated with neoantigen (P=0.95, Figure 5D).
Associations of CDCA7 with immune infiltrations, tumor microenvironment and methyltransferase in ccRCC
Through analyzing the correlation of CDCA7 and six immune cell infiltration levels in ccRCC via online analysis TIMER, we found that in ccRCC, CDCA7 was in close connection with the immune infiltration including B cell infiltration, CD4+ T cell infiltration, CD8+ T cell infiltration, neutrophil infiltration, macrophage infiltration, and dendritic cell infiltration (P＜0.001, Figure 6A). As to the tumor microenvironment, CDCA7 was shown to be involved in immune cells, stromal cells and both of them (P＜0.001, Figure 6B). In addition, the DNA methyltransferase DNMT1 (P<0.001), DNMT2 (P<0.05), DNMT3 (P<0.001), DNMT4 (P<0.001) were also significantly associated with the CDCA7 expression level in ccRCC (Figure 6C).
Associations of CDCA7 with immune checkpoint molecules, immune pathways and mismatch repair proteins
In this study, we analyzed the associations of CDCA7 and 47 immune checkpoint molecules in ccRCC and finally found 25 significantly related molecules including CTLA4, CD274, LAG3 and so on (Figure 7A). We further explored the expression levels of these 25 genes between normal kidney tissues and ccRCC tissues by utilizing the TCGA dataset and eventually identified 10 genes including LAG3, CD27, CD44, CD86, CD276, HHLA2, LAIR1, LGALS9, TIGIT, TNFRSF14 (Figure S1). Besides, relationships between CDCA7 and immune pathways displayed that CDCA7 was closely linked to associated immune cells like activated CD4 T cell, regulatory T cell, memory B cell, macrophage, monocyte and so on (Figure 7B). We also found that CDCA7 was markedly related to mismatch repair proteins including MLH1, MSH2, MSH6, PMS2 in ccRCC (Figure 7C).