Assessment of KAZN mRNA level in ovarian cancer, based on gene expression omnibus (GEO) datasets
The expression data of KAZN in ovarian cancer were obtained through the GEO database. A total of 10 microarrays from the GEO database met the entry criteria. The features of the selected GEO datasets are depicted in Table 1. Of all the data sets, 9 were source tissue samples, one of which was a blood sample. Expression of KAZN was significantly increased in ovarian cancer tissues in GSE105437, GSE18520, GSE27651, GSE36668, GSE38666, GSE40595, GSE66957, GSE69428 (p = 0.0017, p = 0.0012, p < 0.0001, p = 0.0075, p < 0.0001, p < 0.0001, p < 0.0001, p = 0.0177, respectively) (Figure 1). But no statistical difference was detected in GSE29450 (p = 0.0549) and down-regulation of the expression of KAZN was found in normal group in GSE37582 (p < 0.001, supplemental file: Figure S1).
Meta-analysis of GEO datasets
A meta-analysis was conducted based on 10 included microarrays from the GEO database. The results are demonstrated in Figure 2A. Given the apparent heterogeneity (p < 0.01, I2 =91%), a random-effects model was applied, and remarkable up-regulation (SMD = 0.99, 95% CI: 0.12, 1.86) of KAZN mRNA was found in ovarian cancer group.
Sensitivity analysis was performed to explore whether a particular microarray played a vital role in significant heterogeneity (Figure 2B). By removing an individual study per time of meta-analysis to assess the influence of each study, the result showed that no study was found to have played a crucial role in any of the enrolled studies. A funnel plot showed that no evidence of publication bias was observed for this analysis (p = 0.6547, Figure 2C).
Up-regulation of KAZN affects overall survival in ovarian cancer
Since TCGA ovarian cancer datasets have only cancer tissue samples, no normal tissue as a control, so we compare them to GTEx samples, which have expression data from normal ovary tissue of GTEx donors who don’t have cancer. To eliminate the batch effect, we conducted ComBat-seq to integrate different sourced datasets. The comprehensive analysis based on TCGA, GTEx, and GEO datasets showed that the KAZN mRNA expression is significantly differential (p < 0.001) (Figure 3), and the expression was up-regulated in the TCGA group compared to GTEx group (p < 0.0001, Figure 4A).
To further study the clinical effects of KAZN in ovarian cancer, all cases were divided into two groups (KAZN high expressed and KAZN low expressed) by the quantiles of KAZN counts. Then we analyzed the survival status between the two groups. A Kaplan–Meier curve was used to identify the effects of the expression of KAZN on survival time. The result showed that in KAZN low expressed group survival time is significantly longer than in the KAZN high expressed group (Figure 4B). The results based on the Gepia2 (http://gepia2.cancer-pku.cn/#survival) online website analysis confirm our above analysis (supplemental file: Figure S2).
KAZN protein is Up-regulated in ovarian cancer
The Clinical Proteomic Tumor Analysis Consortium (CPTAC) has produced huge amounts of cancer proteomics data providing unprecedented research opportunities. The dataset from CPTAC Ovarian Cancer Confirmatory Study contains 41 normal participants and 169 tumor participants, was used to validate the KAZN protein expression in Ovarian cancer. The result showed that in the tumor group, the KAZN protein expression was significantly higher than in the normal group (p=0.0027, Figure 5). KAZN protein expression was consistent with mRNA level in this study.
KAZN methylation level is correlated to mRNA expression in ovarian cancer
Methylated CpG sites have a moderate/strong association with gene expression changes across the phases in cancer-involved genes with specific functions. To study the correlation between the expression of KAZN and DNA methylation, we detected the DNA methylation level of CpG sites in the KAZN gene body region in TCGA datasets. From the TCGA database, we obtained 27K DNA methylation array data of ovarian cancer which contains 601 tumors and 12 normal samples. 27K methylation array covers only 2 CpG sites for KAZN, respectively is cg21045388 and cg17657618. The cg21045388 is highly methylated in almost all samples. So we detected the correlation between the expression of KAZN and cg17657618. We found the methylated CpG site cg17657618 positively correlated with the rising expression of the KAZN gene (Figure 6A).
To detect the correlation between KAZN DNA methylation and progress of OC, we divided all cases into two groups by stage of clinical traits. We compared the methylation level between the two groups. The result showed that the methylation level of the stage “I and II” group is significantly lower than the stage “III and IV” group (Figure 6B). The results suggested that KAZN methylation may have an important role in the development and outcome of ovarian cancer.
The diagnostic value of KAZN methylation status in ovarian cancer
Illumina Infinium 450K BeadChip covers 134 CpG sites for KAZN. Based on the GSE146552 dataset, we found 13 differentially methylated CpG sites in the KAZN gene body region (details in supplemental table S1). Among these differentially methylated CpG sites, six were hypermethylated and seven were hypomethylated. Interestingly, an unsupervised hierarchical clustering based on the 13 different methylated CpG sites in KAZN almost perfectly divided the samples of GSE146552 into two groups, the EOC (Epithelial ovarian cancer) cluster, and the NE (Not Epithelial OC) cluster (Figure 7). The above results suggest that this CpGs pattern may as a new biomarker contribute to the diagnosis of ovarian cancer.