Identification of genes predicting chemoresistance and short survival in ovarian cancer

Abstract

Background: Ovarian cancer (OC) is a kind of lethiferous cancer in gynecology, and the development of chemoresistance is the brief reason for treatment failure. The genes which contribute to chemoresistance are often leading to short survival. Thus, this study aims to identify predictive markers for chemoresistance and survival from chemoresistant-related genes.

Methods: Coremine was used to retrieve of genes linked to OC chemoresistance. The relationship of genes with patient survival was analyzed in 489 OC patients of The Cancer Genome Atlas (TCGA) cohort, which the subgroup of 90 resistant and 197 sensitive samples was used to determine gene expression. Kaplan-Meier (KM) plotter of 1,816 OC patients with survival data was retrieved for survival analysis. Survival analysis was carried out by the R survival package in R (version 3.3.1). KM and receiver operating characteristic (ROC) curve were respectively used to access the ability of a gene to predict survival and chemoresistance.

Results: In this study, a group of genes potentially linked to OC chemoresistance was identified, which dysregulated in 90 chemoresistant tissues compared with 197 sensitive tissues. Of them, thirteen genes could predict chemoresistance in 1,347 patients, especially SOS1, MSH6, STAT5A were excellent for predicting chemoresistance to any drugs, platin and taxane, CASP2 and PARD6B for any drugs and platin, and HSP90AA1 and HSP90B1 for taxane. Meanwhile, 44 genes linked to OC chemoresistance could predict short overall survival (OS) and/or disease-free survival (DFS) in 489 OC patients, and 10 of them could predict short OS in large cohort of up to 1,657 patients. Finally, it is noteworthy that CASP2 was down-regulated in 90 chemoresistant samples, and low expression of the gene predicted chemoresistance in 1,347 patients, short OS and DFS in 489 patients, and short OS and progression-free survival (PFS) in 1,657 patients.

Conclusions: The identified genes specifically the CASP2 might be potentially used as predictive marker, prognostic marker and therapeutic target in management of OC.

Keywords: Ovarian cancer (OC); chemoresistance; predictive markers.

Figure 1

The 301 genes differentially expressed in 90 chemoresistant samples in contrast to 197 sensitive samples according to TCGA ovarian cancer cohort. (A) Function annotation and enrichment analyses revealed the top 17 biological processes (FDR <1×10⁻⁶) that were differentially expressed, bubbles in the same cluster are represented by the same color; (B) enriched top 10 typical pathways that correlated with ovarian cancer chemoresistance. The X-axis represents the number of genes. TCGA, The Cancer Genome Atlas; FDR, false discovery rate.

Figure 2

Twenty-six genes significantly and differentially expressed in chemoresistant samples in contrast to the sensitive samples, in accordance with TCGA ovarian cancer cohort. Sensitive: 197 platinum sensitive ovarian cancer samples; resistant: 90 resistant samples. *, P<0.05; **, P<0.01. TCGA, The Cancer Genome Atlas.

Figure 3

Role of genes in prediction of chemoresistance in ovarian cancer. Transcriptome-level data of 1,347 ovarian cancer patients were included according to ROC plotter. The relapse status at 6 months was used as a cut off for definition of patient’s response to therapy, and those relapsed within the 6 months were considered as non-responders. (A) Abnormal expression of genes predicts chemoresistance to any drugs (includes platin, taxane, docetaxel, paclitaxel, gemcitabine, topotecan and avastin), in which 130 non-responders and 1,217 responders were included. (B) Abnormal expression of genes predicts chemoresistance to platin, in which 114 non-responders and 1,095 responders were included. (C) Abnormal expression of genes predicts chemoresistance to taxane, in which 81 non-responders and 807 responders were included. AUC, area under the curve; FPR, false positive rate; TPR, true positive rate; ROC, receiver operating characteristic.

Figure 4

Kaplan-Meier analyses determined that seventeen genes were relevant to overall survival and disease-free survival in ovarian cancer, based on TCGA cohort of 489 patients. Gene expression was divided into low (L) and high (H) by the median value. DFS, disease-specific survival; OS, overall survival; TCGA, The Cancer Genome Atlas.

Figure 5

Genes correlated with OS and PFS in 1,816 ovarian cancer patients based on the KM plotters collection. Gene expression was divided into low (L) and high (H) by the median value. (A) Genes related to OS in 489 patients of TCGA cohort consistently predicted OS and/or PFS in 1,816 patients of KM plotter collection. (B) Genes related to DFS in 489 patients was also relevant to OS and/or PFS in 1,816 patients. OS, overall survival; PFS, progression-free survival; KM, Kaplan-Meier; TCGA, The Cancer Genome Atlas.