Expression of KIF18A in tumors and normal tissues
The expression of specific genes in the normal tissues of specific tumors compared with tumor tissues can be analyzed through GEPIA. GEPIA analysis results showed that the expression of KIF18A was significantly increased in many tumor tissues, such as BLCA, BRCA, CESC and so on. Meanwhile, the results demonstrated that the expression of KIF18A increased in GBM (Figure 1a).
In order to further explore the expression of KIF18A in glioma, two GEO datasets (GSE4290 and GSE50161) were enrolled and analyzed. The results demonstrated that the expression of KIF18A in tumor sample tissues up-regulated significantly, compared to normal brain tissue (Figure 1b and c).
Mutiple different sources of data contain 1781 glioma and 36 normal tissues. We found that the expression of KIF18A increased in glioma, but its role in glioma needs further study.
Increased KIF18A expression reduces overall survival of glioma patients
In order to explore the function of KIF18A expression in glioma, we first studied whether KIF18A expression affects the survival of patients with glioma. Kaplan-Meier method was used in this study. A total of 1670 glioma patients had KIF18A expression information from CGGA RNA-seq (n = 749), CGGA microarray (n = 268) and TCGA RNA-seq (n = 653) sets. After analyzing each data set, we found that the overall survival time of patients with higher KIF18A expression was shortened in all grades gliomas (Figure 2, p < 0.001), but the relationship between higher KIF18A expression and survival time was different in WHO grade Ⅱ, WHO grade Ⅲ and WHO grade Ⅳ gliomas. In CGGA RNA-seq data set, for WHO grade Ⅱ (Figure 2 b, p = 0.014) and WHO grade Ⅲ (Figure 2 c, p < 0.001) glioma patients, higher KIF18A expression shortened the survival period, but for WHO grade Ⅳ (Figure 2 d, p = 0.101) glioma patients, there was no significant difference between high and low expression of KIF18A patients. In CGGA microarray and TCGA RNA-seq datasets, higher KIF18A expression decreased the survival time in WHO grade Ⅲ (Fig. 2 g and k, p < 0.01) glioma patients, but had no significant effect on WHO grade Ⅱ (Figure 2 f, p = 0.601 and j, p = 0.66) and WHO grade Ⅳ (Figure 2 h, p = 0.38 and l, p = 0.402) glioma patients. From these data, we found that higher KIF18A expression shortened the overall survival in glioma and the survival period of patients with WHO grade Ⅲ glioma, but the effect on the survival period of patients with WHO grade Ⅱ and WHO grade Ⅳ glioma was not statistically significant.
KIF18A is an independent prognostic factor and increased KIF18A expression is associated with poor prognosis
The high expression of KIF18A can reduce the overall survival of glioma patients. Whether KIF18A is an independent prognostic factor needs to be further verified. In this study, univariate and multivariate regression analysis were used to analyze the three data sets. In CGGA RNA-seq data set, univariate analysis showed that KIF18A (HR = 2.201; 95% CI = 1.954 – 2.479; P < 0.001), prs_type, histology, grade, age, chemo, idh_mutation and 1p19q_codelization could predict the overall survival in all grades gliomas (Figure 3a). In multivariate regression analysis, KIF18A (HR = 1.372; 95% CI = 1.186 – 1.586; P < 0.001) still significantly affected the prognosis with adjusting prs_type, histology, grade, age, chemo, idh_mutation, 1p19q_code selection (Figure 3B). Univariate analysis showed that KIF18A (HR = 1.649; 95% CI = 1.486 – 1.830; P < 0.001), TCGA_subtypes, PRS_type, histology, grade, age and IDH_mutation could predict the overall survival in all grade gliomas in the CGGA microarray dataset (Figure 3c). For multivariate analysis, TCGA﹣subtypes, PRS﹣type, histology, grade, age, and IDH﹣mutation were corrected, and KIF18A (HR = 1.372; 95% CI = 1.203 – 1.566; P < 0.001) was significant (Figure 3d). Univariate analysis in TCGA RNA-seq showed that KIF18A (HR = 1.432; 95% CI = 1.330 – 1.542; P < 0.001), age and grade were significantly associated with overall survival (Figure 3e). In multivariate regression analysis, with age and grade adjusting, KIF18A (HR = 1.224; 95% CI = 1.067 – 1.404; P < 0.001) still had significant effect on prognosis (Figure 3f). Above data show that KIF18A plays an important role in glioma progression and is closely related to poor prognosis.
KIF18A is a biomarker of glioma
To assess KIF18A diagnostic value in glioma, the receiver operating characteristic curve were obtained from three datasets. In CGGA RNA-seq data set, the areas under the curves (AUC) for one year, three year and five year are 0.677, 0.764 and 0.791 respectively. AUC were 0.739, 0.819 and 0.780 respectively in CGGA microarray data set and 0.801, 0.860 and 0.818 respectively in TCGA RNA-seq set. The diagnostic value of KIF18A in different grades gliomas was further analyzed. The results are shown in Figure 4 (b, c, d, f, g, h, j, k and l). These results show that KIF18A has reliable predictive and diagnostic value in glioma.
Relationship between KIF18A expression and clinical characteristics of glioma
The relationship between the KIF18A expression and the clinical characteristics of glioma was analyzed from three data sets. In CGGA RNA-seq data set, the increased expression of KIF18A is closely related to WHO grade, chemo status, age, PRS type, IDH mutation status, 1p19q coding status and histology, as shown in Figure 5 (a, b, c, d, e, f and g). In CGGA microarray data set, the increased KIF18A was significantly related to WHO grade, age, IDH mutation status and history. TCGA RNA-seq data set analysis showed that increased KIF18A expression was significant in high-grade gliomas and in patients aged over 51. These results suggested that increased KIF18A expression is more likely to occur in high-grade, elderly, IDH wide type glioma patients.
Expression of KIF18A protein in glioma
To further explore the expression of KIF18A protein in glioma, immunohistochemical sections of normal brain tissue (Figure 6a), low (Figure 6b), and high grade gliomas (Figure 6c) were downloaded from the human protein atlas. The data showed that the expression of KIF18A in gliomas was enhanced compared to normal brain tissues, and the expression of KIF18A was the most significant in high-grade gliomas.
KIF18A related signaling pathways based on GSEA
Gene set enrichment analysis was used to identify glioma related signaling pathways between low and high KIF18A expression in three data sets. GSEA showed significant differences (FDR < 0.25, NOM p-val < 0.05) in enrichment of MSigDB Collection (c2.cp.biocarta and h.all. v6.1. symbols). We selected four pathways, including the KEGG_DNA_REPLICATION, the KEGG_CELL_CYCLE, the KEGG_MISMATCH_REPAIR and the KEGG_NUCLEOTIDE_EXCISION_REPAIR, showed significantly differential enrichment in KIF18A high expression phenotype based on NES, NOM P value, and FDR value (Figures 6a - d; Table 1), which suggested KIF18A play a special role in the development of glioma.
Table 1.The gene set enriches the high KIF18A expression phenotype.
|
CGGA RNA-seq
|
CGGA microarray
|
TCGA RNA-seq
|
Gene set name
|
NES
|
NOM p-val
|
FDR q-val
|
NES
|
NOM p-val
|
FDR q-val
|
NES
|
NOM p-val
|
FDR q-val
|
KEGG_DNA_REPLICATION
|
1.856
|
0
|
0.154
|
1.880
|
0.006
|
0.0423
|
1.912
|
0
|
0.017
|
KEGG_CELL_CYCLE
|
1.924
|
0.001
|
0.116
|
2.123
|
0
|
0.002
|
2.134
|
0
|
0.005
|
KEGG_MISMATCH_REPAIR
|
1.8019
|
0
|
0.119
|
1.873
|
0.002
|
0.0371
|
1.931
|
0
|
0.016
|
KEGG_NUCLEOTIDE_EXCISION_REPAIR
|
1.774
|
0.001
|
0.114
|
1.764
|
0.009
|
0.093
|
1.954
|
0
|
0.013
|
NES: normalized enrichment score; NOM: nominal; FDR: false discovery rate. Gene sets with NOM P-value <0.05 and FDR q-value <0.25 were considered as significantly enriched.