Meta-analysis of microarray data to determine gene indicators involved in the cisplatin resistance in ovarian cancer

Abstract

Background: Significant miss-expressed gene indicators contributing to cisplatin resistance in ovarian cancer have not been completely understood. It seems that several regulatory genes and signaling pathways are associated with the emergence of the chemo-resistant phenotype.

Aims: Here, a meta-analysis approach was adopted to assess deregulated genes involved in relapse after the first line of chemotherapy (cisplatin).

Methods and results: To do so, six ovarian cancer libraries were gathered from GEO repository. Batch effect removal and quality assessment, and boxplots and PCA were performed using SVA and ggplot2 packages in R, respectively. Cisplatin-resistant and -sensitive ovarian cancer groups were compared with find genes with significant expression changes using linear regression models in the LIMMA R package. The significance threshold for DEGs was taken as adj p-value < .05 and - 1 > logFC > 1. A total of 261 genes were identified to have significant differential expression levels in the cisplatin-resistant versus cisplatin-sensitive group. Among the 10 top up-regulated and down-regulated genes, PITX2, SNCA, and EPHA7 (up), as well as TMEM98 (down) are indirect upstream regulators of PI3K/AKT signaling pathway, contributing greatly to the development of chemo-resistance in cancer via promoting cell proliferation, survival, and cell cycle progression as well as inhibiting apoptosis. Moreover, a comprehensive assessment of DEGs revealed the dysregulation of not only membrane ion channels KCa1.1, Kv4, and CACNB4, affecting cell excitability, proliferation, and apoptosis but also cell adhesion proteins COL4A6, EPHA3, and CD9, affecting the attachment of normal cells to ECM and apoptosis, introducing good options to reverse cisplatin resistance.

Conclusion: Our results predict and suggest that upstream regulators of PI3K/AKT signaling pathway, ion channels, and cell adhesion proteins play important roles in cisplatin resistance development in ovarian cancer.

Keywords: cisplatin resistance; meta-analysis; ovarian cancer.

FIGURE 1

Meta‐analysis workflow of microarray data sets.

FIGURE 2

Boxplot of ovarian cancer sample arrays (A) before and (B) after batch correction. The concentrated distribution of chips around zero indicates that undesired variation has almost been eliminated. R language program was used to develop the plot.

FIGURE 3

PCA plots of data summarized via batch correction. PCA scatter plots give PC1 versus PC2 outputs for each calibrated ovarian cancer sample of data sets (A) before and (B) after batch correction. PCA plots developed using ggplot2 package version 3.3.3 in R indicate that gene expression profiles of samples are similar.

FIGURE 4

Volcano plots of DEGs. Ggplot2 package in R74 was used to visualize DEGs discovered via meta‐analysis. Cut‐off for p‐value was considered .05. It is seen that 30 genes were revealed as common DEGs with a change greater than two folds in chemo‐resistant relative to chemo‐sensitive group. Data are provided as log2 fold change. Ggplot2 R package version 3.3.363 was utilized to create the plots.

FIGURE 5

Interaction between genes that are up‐regulated in ovarian cancer cells, cisplatin resistance cell line, A2780CP, relative to cisplatin‐sensitive cell line, A2780S, according to DEGs. Darker color and bigger circles represents higher degree.

FIGURE 6

Interaction between genes that are down‐regulated in ovarian cancer cells, cisplatin resistance cell line, A2780CP, relative to cisplatin‐sensitive cell line, A2780S according to DEGs. Darker color and bigger circles represents higher degree.

FIGURE 7

Heatmap of Log fold variations for DEGs with most significant regulation level. Heatmap indicates the normalized relative expression values of DEGs with most significant regulation level involved in drug resistance between cisplatin‐resistant and cisplatin‐sensitive ovarian samples. Each column indicates ovarian samples. All the samples were separated into two clusters according to their characteristics (resistant and sensitive to chemotherapy). Higher gene expression based on microarray data shows with darker color.