Abstract
BACKGROUND: It has been demonstrated by studies globally that RNA binding proteins (RBPs) took part in the development of cervical cancer (CC). Few studies concentrated on the correlation between RBPs and overall survival of CC patients. We retrieved significant DEGs (differently expressed genes, RNA binding proteins) correlated to the process of cervical cancer development. METHODS: Expressions level of genes in cervical cancer and normal tissue samples were obtained from GTEx and TCGA database. Differently expressed RNA binding proteins (DEGs) were retrieved by Wilcoxon sum-rank test. ClusterProfiler package worked in R software was used to perform GO and KEGG enrichment analyses. Univariate proportional hazard cox regression and multivariate proportional hazard cox regressions were applied to identify DEGs equipped with prognostic value and other clinical independent risk factors. ROC curve was drawn for comparing the survival predict feasibility of risk score with other risk factors in CC patients. Nomogram was drawn to exhibit the prediction model and validated by C-index and calibration curve. Correlations between differentially expressed RNA binding proteins (DEGs) and other clinical features were investigated by t test or Cruskal Wallis analysis. Correlation between Immune and DEGs in cervical cancer was investigated by ssGSEA. RESULTS: 347 differentially expressed RBPs (DEGs) were retrieved from cervical cancer tissue and normal tissue samples. GO enrichment analysis showed that these DEGs involved in RNA splicing, catabolic process and metabolism. Cox regression model showed that there were ten DEGs significantly associated with overall survival of cervical cancer patients. WDR43 (HR = 0.423, P = 0.008), RBM38 (HR = 0.533, P < 0.001), RNASEH2A (HR = 0.474, P = 0.002) and HENMT1 (HR = 0.720, P = 0.071) played protective roles in survival among these ten genes. Stage (Stage IV vs Stage I HR = 3.434, P < 0.001) and risk score (HR = 1.214, P < 0.001) were sorted as independent prognostic risk factors based on multivariate cox regression. ROC curve validated that risk score was preferable to predict survival of CC patients than other risk factors. Additionally, we found some of these ten predictor DEGs were correlated significantly in statistic with tumor grade or stage, clinical T stage, clinical N stage, pathology or risk score (all P < 0.05). Part of immune cells and immune functions showed a lower activity in high risk group than low risk group which is stratified by median risk score. CONCLUSION: Our discovery showed that many RNA binding proteins involved in the progress of cervical cancer, which could probably serve as prognostic biomarkers and accelerate the discovery of treatment targets for CC patients.