Identification of exosomes-related lncRNAs in clear cell renal cell carcinoma based on Bayesian spike-and-slab lasso approach.

Exosomes-related long non-coding RNAs (lncRNAs) have been reported to play significant roles in clear cell renal cell carcinoma (ccRCC). However, there is little known about the relationship between exosomes-related lncRNAs and ccRCC. This study aimed to select optimal prognostic model based on exosomes-related lncRNAs to provide a methodological reference for high-dimensional data. Based on the Cancer Genome Atlas (TCGA) database of 515 ccRCC patients, two risk score models were generated underlying Bayesian spike-and-slab lasso and lasso regression. The optimal model was determined by calculating the area of time-dependent receiver-operating characteristic (ROC) curves in the TCGA and ArrayExpress databases. The immune patterns and sensitivity of immunotherapy between the high and low groups were further explored. Initially, we constructed two risk score models containing 11 and 7 exosomes-related lncRNAs according to Bayesian spike-and-slab lasso and lasso regression respectively. ROC curves revealed that the model constructed by Bayesian spike-and-slab lasso regression was more reliable in predicting survival at 1, 3, and 5 years, yielding an area under the curves (AUCs) of 0.796, 0.732, and 0.742, respectively. Kaplan-Meier (K-M) curves presented that prognosis was poorer in the high-risk score group (P < 0.001). Additionally, the high-risk score group patients were enriched in immune-activating phenotypes and more sensitive to immunotherapy. The exosomes-related lncRNAs model constructed with Bayesian spike-and-slab lasso regression has higher predictive power for ccRCC patients' prognosis, which provides methodological reference for the analysis of high-dimensional data in bioinformatics and guides the tailored treatment of ccRCC patients.