In silico analyses identify gene-sets, associated with clinical outcome in ovarian cancer: role of mitotic kinases

Abstract

Introduction: Accurate assessment of prognosis in early stage ovarian cancer is challenging resulting in suboptimal selection of patients for adjuvant therapy. The identification of predictive markers for cytotoxic chemotherapy is therefore highly desirable. Protein kinases are important components in oncogenic transformation and those relating to cell cycle and mitosis control may allow for identification of high-risk early stage ovarian tumors.

Methods: Genes with differential expression in ovarian surface epithelia (OSE) and ovarian cancer epithelial cells (CEPIs) were identified from public datasets and analyzed with dChip software. Progression-free (PFS) and overall survival (OS) associated with these genes in stage I/II and late stage ovarian cancer was explored using the Kaplan Meier Plotter online tool.

Results: Of 2925 transcripts associated with modified expression in CEPIs compared to OSE, 66 genes coded for upregulated protein kinases. Expression of 9 of these genes (CDC28, CHK1, NIMA, Aurora kinase A, Aurora kinase B, BUB1, BUB1βB, CDKN2A and TTK) was associated with worse PFS (HR:3.40, log rank p<0.001). The combined analyses of CHK1, CDKN2A, AURKA, AURKB, TTK and NEK2 showed the highest magnitude of association with PFS (HR:4.62, log rank p<0.001). Expression of AURKB predicted detrimental OS in stage I/II ovarian cancer better than all other combinations

Conclusion: Genes linked to cell cycle control are associated with worse outcome in early stage ovarian cancer. Incorporation of these biomarkers in clinical studies may help in the identification of patients at high risk of relapse for whom optimizing adjuvant therapeutic strategies is needed.