PANK1 expression is correlated with ccRCC clinicopathological features
To identify the difference in PANK1 expression, we analyzed PANK1 expression levels in 539 ccRCC tissues and 72 adjacent normal renal tissues and we found low expression of PANK1 in ccRCC tissues (P<0.001, Figure 1A). At the same time, we also analyzed the expression of PANK1 in 72 ccRCC tissues and their matched neighboring tissues. The results showed low expression of PANK1 in ccRCC tissues (P<0.001, Figure 1B). Meanwhile, the expression levels of PANK1 in the normal samples from the GTEx combined TCGA database and the ccRCC samples from the TCGA database were compared. To determine the differential expression of PAK1 in tumor and normal tissues, transcriptional levels of PAK1 in different multiple cancer types and normal tissues were analyzed using TCGA and GTEx databases. This analysis showed higher expression of PANK1 in various types of cancer than in normal tissues (Figure S1). We downloaded the unified and standardized pan-cancer data set: TCGA Target GTEX (Pancan, n = 19131, g = 60499) from the UCSC(https://xenabrowser.net/) database, and further extracted the expression data of PANK1 gene in each sample. Further, we screened the sample sources as follows: Solid Normal, Primary Solid Tumor, Primary Tumor, Normal Tissue, Primary Blood Derived Cancer-Bone Marrow, The samples of Primary Blood Derived Cancer-Peripheral Blood are further subjected to log2(x+0.001) transformation for each expression value. Finally, we also removed the cancer species with the number of samples less than three from a single cancer species, and finally obtained the expression data of 34 cancer species. We used the R software (version 3.6.3) to calculate the expression differences of normal samples and tumor samples in each tumor, and performed difference significance analysis using unpaired Wilcoxon Rank Sum and Signed Rank Tests. We observed significant up-regulation in 19 tumors such as GBMLGG(Tumor:2.15±0.99,Normal:1.31±1.50,p=4.2e-47)、LGG(Tumor:2.39±0.90,Normal:1.31±1.50,p=1.5e-61)、UCEC(Tumor:2.09±1.05,Normal:0.82±0.70,p=4.5e-8)、BRCA(Tumor:1.38±1.33,Normal:0.75±0.83,p=2.7e-20)、CESC(Tumor:1.76±1.08,Normal:1.21±0.70,p=0.02)、LUAD(Tumor:1.59±0.98,Normal:0.78±1.15,p=1.0e-25)、ESCA(Tumor:1.96±1.11,Normal:0.90±1.32,p=1.3e-25)、STES(Tumor:2.28±1.07,Normal:0.94±1.33,p=2.6e-94)、COAD(Tumor:3.29±0.82,Normal:1.39±2.04,p=5.7e-42)、COADREAD(Tumor:3.34±0.79,Normal:1.45±2.05,p=2.9e-48)、STAD(Tumor:2.42±1.02,Normal:1.07±1.38,p=2.7e-43)、LUSC(Tumor:1.50±0.96,Normal:0.78±1.15,p=2.7e-22)、BLCA(Tumor:1.49±1.04,Normal:0.90±0.78,p=1.8e-3)、OV(Tumor:2.11±1.24,Normal:0.91±0.75,p=9.5e-25)、PAAD(Tumor:1.14±0.79,Normal:-0.52±1.24,p=2.1e-45)、TGCT(Tumor:1.90±1.08,Normal:1.72±0.65,p=0.02)、UCS(Tumor:2.28±0.94,Normal:0.71±0.72,p=3.3e-16)、ALL(Tumor:-2.17±2.23,Normal:-4.53±2.20,p=7.7e-26)、LAML(Tumor:1.15±1.22,Normal:-4.53±2.20,p=2.3e-74) and we observed significant downregulation in 10 tumors such as KIRP(Tumor:2.27±1.10,Normal:3.98±1.68,p=1.6e-38)、KIPAN(Tumor:2.61±1.32,Normal:3.98±1.68,p=1.1e-36)、HNSC(Tumor:1.13±1.04,Normal:1.61±0.94,p=2.6e-3)、KIRC(Tumor:2.75±1.41,Normal:3.98±1.68,p=7.5e-28)、LIHC(Tumor:3.73±1.05,Normal:4.11±1.10,p=1.5e-6)、WT(Tumor:3.21±0.84,Normal:3.98±1.68,p=9.3e-15)、SKCM(Tumor:0.91±1.14,Normal:2.90±0.77,p=3.2e-45)、THCA(Tumor:2.19±0.98,Normal:2.50±1.07,p=3.2e-8)、KICH(Tumor:2.96±1.04,Normal:3.98±1.68,p=1.4e-11)、CHOL(Tumor:2.46±1.12,Normal:4.94±0.46,p=2.3e-9) (Fig. S1). In addition, the receiver operating characteristic (ROC) curve was used to analyze the effectiveness of ccRCC expression levels in distinguishing ccRCC tissues from non-tumor tissues. The area under the curve (AUC) of PANK1 was 0.898, indicating that PANK1 could be an ideal biomarker to differentiate ccRCC from non-neoplastic tissues (Figure 1C).
Patient characteristics are presented in Table 1, with 539 cases of primary ccRCC with clinical and gene expression data collected from the TCGA database. According to the average relative expression level of PANK1, patients with ccRCC were divided into high-expression group (n=270) and low-expression group (n=269). To evaluate the correlation between the expression of PANK1 and the clinical pathological features of patients with ccRCC. The chi-square test showed that the expression of PANK1 was related to gender (P<0.001), T stage (P<0.001), histological grade (P<0.001), pathological stage (P<0.001), N stage (P<0.05), and M stage (P<0.05).
Logistic regression was used to analyze the relationship between the clinical pathological characteristics of ccRCC and the expression level of PANK1.
The results suggested that PANK1 had a significant correlation with gender (P<0.001), T stage (P<0.001), histological grade (P<0.001), pathological stage (P<0.001), N stage (P=0.012), and M stage (P=0.007) (Table 2, Figure 2).
PANK1 expression associated with poor prognosis in ccRCC patients
The association of PANK1 expression with PFS in ccRCC patients was assessed by Kaplan-Meier analysis and showed a negative association of PANK1 expression with poor OS in ccRCC patients (P<0.001, Figure 3A). In addition, to expand our observation to pan-cancer levels, the relationship between expression of PANK1 and patient survival was further analyzed in a variety of cancer types other than ccRCC. As shown in Figure S2, a significant association between PANK1 expression and poor OS was also observed in patients with colon cancer (COAD), renal papillary cell carcinoma (KIRP), brain low-grade gliomas (LGG), mesothelioma (MESO), pancreatic cancer (PAAD), and rectal adenocarcinoma (READ).
Cox univariate and multivariate analysis of prognostic factors in ccRCC
Table 3 shows the Cox univariate and multivariate analysis results for OS in patients with ccRCC. P<0.01 variables in the Cox univariate regression model were age (P<0.001), T stage (P<0.001), N stage (P<0.001), M stage (P<0.001), pathological stage (P<0.001), histological grade (P<0.001), and PANK1 (P<0.001). Multivariate analysis further revealed that age (P=0.023), M stage (P<0.001), pathological stage (P=0.042), histological grade (P=0.014), and PANK1 (P<0.001) were independent prognostic factors for OS in patients with ccRCC(Table3,Figure4). A nomogram ,was developed, that is able to predict 1,3 and 5-year OS using the PANK1 gene and other clinical features of KIRC (including age, gender, T stage, N stage, M stage, pathological stage, and histologic grade). To read the nomogram, a vertical line up to the top point row to assign points for each variable should be drawn. Then, the total points for a patient can be added up, and one can obtain the probability of 1,3 and 5-year OS by drawing a vertical line from the total points row. The calibration plots for the probabilities of 1,3 and 5-year OS showed good agreement between the predicted OS by nomogram and actual OS of ccRCC patients (Figure5).
Correlation signal path of PANK1 based on GSEA
The KEGG signaling pathway associated with PANK1 was identified using the GSEA method. GSEA showed significant differences in MSigDB enrichment (c5) (Padj <0.05, FDR <0.25). According to the screening conditions, we screened four significantly related signaling pathways from the KEGG signaling pathway enriched in GSEA, namely, the extracellular matrix receptor pathway, the signaling pathway related to hypertrophic cardiomyopathy, the cytokine-cytokine receptor interaction pathway, and the complement and coagulation cascade pathway (Table 4, Figure6 )。
Correlation between expression of PANK1 and immune infiltration
We further analyzed the correlation between the expression of PANK1 and the immune invasion of ssGSEA using Spearman R. The results showed that the expression of PANK1 was negatively correlated with the infiltration levels of aCD, B cells, CD8+T cells, cytotoxic cells, macrophages, natural killer (NK)CD56 bright cells, plasma cell-like dendritic cells (pDC), T cells, Tem cells, Th1 cells, Th2 cells, and Treg cells (P<0.001), and positively correlated with the infiltration levels of eosinophils, neutrophils, and Th17 cells (P<0.001). (Table5, Figure7 and FigureS3)