Ovarian Cancer Prognosis Analysis Based On Multi-dimensional Genomics And Pathway Activity

Ovarian cancer is one of the most common gynecologic malignancies and has the highest case-fatality rate among gynecologic cancers.The outcome of ovarian cancer was poor and varied greatly among patients,and the molecular mechanisms underlying ovarian cancer prognosis remain unclear.With the implementation of large-scale cancer genome project,it has accumulated abundant genome-wide and clinical data of cancer samples and provides an unprecedented opportunity for a comprehensive interpretation of the molecular mechanisms of cancer development.Because the high-throughput data typically is highly variable and correlated,moreover,ovarian cancer is a complex disease with high molecular heterogeneity,which has extremely complex and diverse changes on the single gene level.Therefore,it is reasonable to explore molecular mechanisms underlying cancer occurrence and development from the perspective of biological pathways or function modules,rather than just on the single gene level.This research explored prognosis-related biological mechanisms of ovarian cancer from the aspects of the molecular subtypes and prognostic prediction based on functional pathways,in order to provide new indicators for individualized treatment and a new strategy for targeted cancer therapy.This thesis specifically includes the following parts:(1)By the integration of multi-dimensional genomics data,pathways were enriched by the prognosis related genes of interest and then ovarian cancer subtypes were distinguished based on these enriched pathway features.Through gene expression profiling,combined with copy number variation,DNA methylation and miRNA expression,prognosis-related genes of ovarian cancer were selected using the Cox regression model.Based on pathways which were enriched by the prognosis related genes,ovarian cancer subtypes were distinguished by estimating pathway activity.Further,the survival characteristics of cancer subtypes were analyzed to establish the relationship between prognosis of subtypes and regulation pattern of pathways.Also,the differences between the cancer subtypes identified by our method and clinical stage were compared.Finally the survival and pathway activity characteristics were verified in a large independent validation data set.(2)A prediction model was constructed based on the collaborative pathways which were determined by prognosis-related gene pairs and then the indicators were identified for survival prediction from the model.Through the survival analysis of gene expression profile,prognosis-related gene pairs were selected by regarding the difference between the pair genes as the independent variable in survival analysis.Genes and pathways related to pro-or anti-mortality risk were obtained according to whether the high expression of each gene would increase or decrease mortality risk.For each two pathways,collaborative pathways were determined by judging whether pro-and anti-mortality risk genes from the same prognosis-related pairs were significantly enriched between the two pathways.Further,an optimal model was constructed for survival prediction based on collaborative pathways,and it was compared with the prediction model that was constructed directly based on genes or pathways.Pathway indicators were then extracted from the prediction model and their influence on survival of ovarian cancer patients was analyzed.From the analyses of the molecular subtypes and prognostic prediction,the relationship between prognosis-related pathway activity and survival time could be found and the influence of collaborative pathways on survival of ovarian cancer patients could also be verified.It is not only of valuable reference to the treatment and prognosis of ovarian cancer,but also meaningful to explore the mechanisms underlying cancer occurrence and development.