A Study On The Relationship Between Mi Rna And Prognosis Of High-grade Ovarian Serous Cystadenocarcinoma By Using Artificial Neural Network

Background and purpose: Ovarian cancer(OC)is one of the common gynecological oncology,and the mortality from OC is highest in gynecologic malignant tumor.About 90% of OC are epithelial ovarian cancer(EOC),and ovarian serous cystadenocarcinoma(OV)is the main histological type of EOC(Accounting for 75-80%).Patients with early stages of OV have a good treatment response that most can be cured,however most of patients in advanced stages more than stage II at the first diagnosis can’t avoid death for relapse and metastasis even after giving cytoreductive surgery and postoperative adjuvant chemotherapy actively.Therefore,many researchers are trying to explore specific prognostic factors of ovarian cancer which can guide doctors to take interventions and improve the prognosis initially in clinical practice.Among those studies on prognostic factors,the study about the relationship between micro RNA(miRNA)and ovarian cancer is a prevalent theme recently.Most relevant articles have found that miRNA is associated with the prognosis of ovarian cancer.Mi RNAs are endogenous,non-coding RNAs,which are highly conserved in vivo and posttranscriptional regulators of gene expression.Mi RNAs play complex roles in several different aspects of many biological processes.Almost all of the previous studies assume a linear relationship between miRNA expression and disease phenotype,but this assumption is still not confirmed.Numerous studies have shown that a miRNA can regulate multiple messenger RNA(m RNA)while several miRNAs can regulate one m RNA precisely.In other words,miRNAs are not independent of each other.In other words,miRNAs are not independent of each other.Thus,we speculate that the relationship between miRNA expression and diseasephenotype is non-linear rather than linear.Artificial neural network(ANN)is an artificial intelligent model.ANN is an information processing structure similar to a biological neural network,which has a good predictive performance when solves non-linear,unstable,and complex problems.And that multi-layer perceptron(MLP)is a common and simple feedforward neural network similar to a biological neuron.ANN is a potential and powerful tool for bioinformatics that has been used in many areas and different levels successfully.Some scholars have counted the number of Bioinformatics papers related to ANNs on the Pub Med website which showed a significant growth trend.The Cancer Genome Atlas(TCGA)project was initiated by the American government.TCGA project is implemented by the National Cancer Institute(NCI)and the National Human Genome Research Institute(NHGRI).TCGA project is the largest genetic engineering researching the changes of the cancer genome in the world currently based on the results of the Human Genome Project(HGP)and is equivalent to more than 100 HGP.TCGA website is a public database based on the results of TCGA project,and it is the largest and the most common public resources that the researchers all over the world can use it freely.Genomic Data Commons Data Portal(GDC)is a powerful data-driven platform of TCGA website,which can link to external analysis tools,such as c Bio Portal website、Firehose website.This study also used these two analysis tools.TCGA researchers have generated,analyzed,and provided the data of genomic sequence,expression,methylation and copy number variation of tumor samples from about 30 kinds of tumors by large-scale genome sequencing and integral multidimensional analysis.In2011,TCGA project has analyzed 488 samples of high-grade ovarian serous cystadenocarcinoma(HGS-OvCa),including their miRNA expression.But,even if TCGA is a powerful,well-organized molecular database,its data mining is still limited.This study includes clinical data and miRNA expression data ofHGS-OvCa from TCGA website,and use ANN model to explore the relationship between miRNA and HGS-OvCa prognosis.Data and Methods: Standard statistical tests were used to analyze the clinical data of HGS-OvCa samples downloading in TCGA website including,but not limited to,log-rank text,and Cox proportional hazard analysis,descriptive statistics,as appropriate.Combined with all statistical analysis results,when p <0.05,it was considered statistically significant.Download miRNAs expression data of HGS-OvCa samples in TCGA website.GDC analysis tools were used to select miRNAs related to the overall survival of HGS-OvCa samples.10 miRNAs that have minimum log-rank p value were used to build the ANN model to predict the prognosis of HGS-OvCa.SPSS19.0 software is the main analytical tools in this study.Results: 1,Age at diagnosis,FIGO stage,residual disease after initial surgical cytoreduction,platinum status was associated with the overall survival of HGS-OvCa samples.2,Mi RNAs expression is different in different HGS-OvCa samples.3,56 miRNAs are related to the overall survival of HGS-OvCa simples selected by using GDC analysis tools in website TCGA and 10 miRNAs were used to build the ANN model to predict the prognosis of HGS-OvCa.Conclusion: 1,Age at diagnosis,FIGO stage,residual disease after initial surgical cytoreduction,platinum status are Clinical prognostic factors of HGS-OvCa.2,Mi RNA expression is related to the prognosis of HGS-OvCa.3,Using miRNA to build ANN model can predict the prognosis of patients with HGS-OvCa.