A novel immune-related lncRNA pair signature for prognostic prediction and immune response evaluation in gastric cancer: a bioinformatics and biological validation study

Abstract

Background Gastric cancer (GC), the most commonly diagnosed cancer worldwide with poor 5-year survival rate in advanced stages. Although immune-related and survival-related biomarkers, which typically comprise aberrantly expressed long non-coding RNAs (lncRNAs) and genes, have been identified, there are no reports of immune-related lncRNA pair (IRLP) signatures for GC. Methods In this study, we acquired lncRNA expression profiles from The Cancer Genome Atlas (TCGA) and used the least absolute shrinkage and selection operator (LASSO) Cox proportional hazards model (iteration = 1000) to develop a IRLP prognostic signature. The area under curve (AUC) was used to assess the prognosis predictive power. The multivariate Cox regression analysis was performed to identify whether this signature was an independent prognostic factor. The immune cell infiltration analysis was performed between the two risk groups. Last, molecular experiments were performed to explore LINC01082 is involved in the development of GC. Results We acquired lncRNA expression profiles and used the LASSO Cox model to develop an 18-IRLP signature with a strong prognostic predictive power. The 5-year AUC values of the training, validation, and overall TCGA datasets were 0.77, 0.86, and 0.80, respectively. The different prognostic outcomes between the high- and low-risk groups were determined using our 18-IRLP signature. Moreover, our 18-IRLP signature was an independent prognostic factor as per the multivariate Cox regression analysis, and showed better prognostic evaluation than the traditional TNM staging system as well as other clinical features. We also found differences in cancer-associated fibroblast and macrophage M2 infiltration and the expression of PD-L1, CTLA4, LAG3, and HLA were also observed between the two risk groups (P < 0.05). Analysis of biological functions revealed that target genes of the lncRNAs in the IRLP signature were enriched in focal adhesion and regulation of actin cytoskeleton. Finally, as one of significant candidates of IRLP signature, overexpression of LINC01082 suppressed the invasion ability of GC cells as well as PD-L1 expression profiles. Conclusions Our novel 18-IRLP signature provides new insights regarding immunological biomarkers, imparts a better understanding of the tumor immune microenvironment, and can be used for predicting prognosis and evaluating immune response in GC.

Critical Analysis of Risk Factors and Machine-Learning-Based Gastric Cancer Risk Prediction Models: A Systematic Review

“…The schematic diagram of random forests is shown in Figure 6B. In support of the finding that the polygenic risk score (PRS) model of six SNPs is capable of predicting the risk of GC, RF analyses demonstrated that the combination of the six SNPs has a high predictive power for GC, with an AUC value of 0.75, which also verifies the high fitting ability of RF (Xiaoyu Wang et al (2022) [23]). Among the current algorithms, random forests have excellent accuracy and fast training speed and can evaluate the importance of features.…”

“…Its mathematical expression is shown as Formula (1). Jun Wang et al (2022) [48] acquired lncRNA expression profiles from TCGA and used the LASSO to develop an immune-related lncRNA pair (IRLP) prognostic signature termed the 18-IRLP signature, which provided new insights regarding immunological biomarkers and could be used for predicting prognosis and evaluating the immune response in GC. The LASSO can be used not only for dimension reduction, but also for regression.…”

Screening and identification of hub genes for ischemic cardiomyopathy and construction and validation of a clinical prognosis model using bioinformatics analysis

“…The research methods mainly include the collection (collection and screening), processing (editing, sorting, management, and display), utilization (calculation and simulation), and analysis of biological data ( 9 , 10 ). Experience has shown that bioinformatics technology has great application value in the screening of disease biomarkers, which is of great significance for the diagnosis, treatment, and prognosis of diseases, and can provide a more comprehensive and profound understanding of diseases ( 11 - 14 ). A recent study showed that differentially expressed genes (DEGs) were significantly enriched in metabolic pathways, oxidative phosphorylation, and extracellular matrix (ECM) receptor interactions, and were closely related to fibrosis, collagen catabolic process and inflammatory response function ( 15 ).…”

Long Non-Coding RNA LOC339059 Attenuates IL-6/STAT3-Signaling-Mediated PDL1 Expression and Macrophage M2 Polarization by Interacting with c-Myc in Gastric Cancer

“…Huang et al reported 23 m6A-related lncRNAs related to tumor-infiltrating immune cells, the expression of programmed death-1 (PD-1), and cytotoxic T-lymphocyteassociated protein 4 (CTLA4) [18]. Wang et al reported eighteen immune-related lncRNA pair (IRLP) signatures associated with cancer-associated fibroblasts; macrophage M2 infiltration; and PD-L1, CTLA4, LAG3 (Lymphocyte Activation Gene-3), and HLA expression in GC [19]. Wang et al reported the expression profiles of 15 m6A-related lncRNAs linked to the immune score, stromal score, and ESTIMATE score in GC [20].…”