Identification of muscle-invasive related genes in bladder cancer single-cell sequencing data for constructing patient prognostic model

Abstract

Abstract Single-cell sequencing is an emerging sequencing technology that can effectively identify the cell types of tumors. In bladder cancer prognosis, muscular invasion often represents a poor prognosis and affects patients' quality of life. This study aims to extract the expression levels of muscle-invasive related genes(MIRGs) in bladder cancer patients and construct a model of MIRG, which can predict bladder cancer patients' prognosis using bioinformatics methods. Methods: Single-cell sequencing data of bladder cancer patients were obtained from the GEO database. After conducting quality control and cell type identification, all epithelial cells in the samples were extracted and classified based on their invasive and non-invasive characteristics, followed by a differential analysis. The results were identified as MIRGs. Subsequently, we downloaded and organized gene data of bladder cancer patients from TCGA and determined the intersection of MIRGs and the sequenced gene set of TCGA patients. Clinical information was then associated with the intersection, and the data were divided into training and test sets, with the training set used for model construction and the test set for model verification. Subsequently, the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm and Cox regression were used to construct a prognostic model based on MIRGs. Based on the prognostic features, risk scores were calculated, and patients were classified into high-risk and low-risk groups. We observed the survival information of patients in the high-risk and low-risk groups in both the training and test sets, constructed ROC curves to assess the predictive ability of the model, and subsequently, we generated nomograms. Results: Three cell types were identified, and epithelial cells were extracted, clustered, and divided into invasive and non-invasive groups based on pathological staging. A total of 411 differentially expressed genes were screened. GO and KEGG analyses revealed that these genes were significantly associated with cellular processes such as apoptosis, cell adhesion, and tumor development and progression.After intersecting the expressed genes, 402 genes were determined for model construction. Following the LASSO algorithm and Cox regression, a risk prediction model consisting of CD74, AKR1B1, EIF3D, EMP1, CRABP2, TRIM31, RPL36A and MRPS6 was established.Survival curves and Receiver Operating Characteristic (ROC) curves demonstrated that the model exhibited good predictive ability. A nomograms was constructed to predict patients' survival rates at 1, 3, and 5 years. The calibration curve of the nomograms indicated that it had a satisfactory prognostic ability for patients. Conclusion: In this study, based on single-cell sequencing data, TCGA sequencing data and clinical information, the bladder cancer muscle-invasive related gene prognostic model constructed using multi-omics methods demonstrated a certain degree of accuracy and reliability in predicting the survival prognosis of bladder cancer patients. This provides a reference for assessing the prognosis of bladder cancer patients.