Combining single-cell sequencing data to construct a prognostic signature to predict survival, immune microenvironment, and immunotherapy response in gastric cancer patients

TL;DR: Developing an immune-related genetic signature at the single-cell level for categorizing GC cases and predicting patient prognostic outcome, immune status as well as treatment response, and the immunotherapeutic response prediction accuracy was validated in an external dataset IMvigor210 cohort.

Abstract: Background and objective Gastric cancer (GC) represents a major factor inducing global cancer-associated deaths, but specific biomarkers and therapeutic targets for GC are lacking at present. Therefore, the present work focused on developing an immune-related genetic signature at the single-cell level for categorizing GC cases and predicting patient prognostic outcome, immune status as well as treatment response. Methods Single-cell RNA-sequencing (scRNA-seq) data were combined with bulk RNA-seq data in GC patients for subsequent analyses. Differences in overall survival (OS), genomic alterations, immune status, together with estimated immunotherapeutic outcomes were measured between different groups. Results Nine cell types were identified by analyzing scRNA-seq data from GC patients, and marker genes of immune cells were also selected for subsequent analysis. In addition, an immune-related signature was established to predict OS while validating the prediction power for GC patients. Afterwards, a nomogram with high accuracy was constructed for improving our constructed signature’s clinical utility. The low-risk group was featured by high tumor mutation burden (TMB), increased immune activation, and microsatellite instability-high (MSI-H), which were related to the prolonged OS and used in immunotherapy. By contrast, high-risk group was associated with microsatellite stability (MSS), low TMB and immunosuppression, which might be more suitable for targeted therapy. Meanwhile, the risk score generated by our signature was markedly related to the cancer stem cell (CSC) index. In addition, the immunotherapeutic response prediction accuracy of our signature was validated in an external dataset IMvigor210 cohort. Conclusion A signature was constructed according to scRNA-seq data analysis. The signature-screened low- and high-risk patients had different prognoses, immune statuses and enriched functions and pathways. Such results shed more lights on immune status of GC, prognosis assessment, and development of efficient immunotherapeutic treatments.