Search

Background: The heterogeneity limits the effective application of immune checkpoint inhibitors for patients with stomach adenocarcinoma (STAD). Precise immunotyping can help select people who may benefit from immunotherapy and guide postoperative management by describing the characteristics of tumor microenvironment. Methods: Gene expression profiles and clinical information of patients were collected from ACRG and TCGA-STAD datasets. The immune subtypes (ISs) were identified by consensus clustering analysis. The tumor immune microenvironments (TIME) of each IS were characterized using a series of immunogenomics methods and further confirmed by multiplex immunohistochemistry (mIHC) staining in clinical samples. Two online datasets and one in-house dataset were utilized to construct and validate a prognostic immune-related gene (IRG) signature. Results: STAD patients were stratified into five reproducible ISs. IS1 (immune deserve subtype) had low immune infiltration and the highest degree of HER2 gene mutation. With abundant CD8+ T cells infiltration and activated cytotoxicity reaction, patients in the IS2 (immune-activated subtype) had the best overall survival (OS). IS3 and IS4 subtypes were both in the reactive stroma state and indicated the worst prognosis. However, IS3 (immune-inhibited subtype) was characterized by enrichment of FAP+ fibroblasts and upregulated TGF-β signaling pathway, while IS4 (activated stroma subtype) was characterized by enrichment of ACTA2+ fibroblasts. In addition, mIHC staining confirmed that TGF-β upregulated FAP+ fibroblasts were independent risk factor of OS. IS5 (chronic inflammation subtype) displayed moderate immune cells infiltration and had a relatively good survival. Lastly, we developed a nine-IRG signature model with a robust performance on overall survival prognostication. Conclusions: The immunotyping is indicative for characterize the TIME heterogeneity and the prediction of tumor prognosis for STADs, which may provide valuable stratification for the design of future immunotherapy. [ABSTRACT FROM AUTHOR]

Key message: Using genomic structural equation modelling, this research demonstrates an efficient way to identify genetically correlating traits and provides an effective proxy for multi-trait selection to consider the joint genetic architecture of multiple interacting traits in crop breeding. Breeding crop cultivars with optimal value across multiple traits has been a challenge, as traits may negatively correlate due to pleiotropy or genetic linkage. For example, grain yield and grain protein content correlate negatively with each other in cereal crops. Future crop breeding needs to be based on practical yet accurate evaluation and effective selection of beneficial trait to retain genes with the best agronomic score for multiple traits. Here, we test the framework of whole-system-based approach using structural equation modelling (SEM) to investigate how one trait affects others to guide the optimal selection of a combination of agronomically important traits. Using ten traits and genome-wide SNP profiles from a worldwide barley panel and SEM analysis, we revealed a network of interacting traits, in which tiller number contributes positively to both grain yield and protein content; we further identified common genetic factors affecting multiple traits in the network of interaction. Our method demonstrates an efficient way to identify genetically correlating traits and underlying pleiotropic genetic factors and provides an effective proxy for multi-trait selection within a whole-system framework that considers the joint genetic architecture of multiple interacting traits in crop breeding. Our findings suggest the promise of a whole-system approach to overcome challenges such as the negative correlation of grain yield and protein content to facilitating quantitative and objective breeding decisions in future crop breeding. [ABSTRACT FROM AUTHOR]