TL;DR
This study introduces a novel prognostic model for gastric cancer (GC) based on pairs of immune-related long noncoding RNAs (IRlncRNAPs). This model accurately predicts patient survival, immune infiltration patterns, and potential response to chemotherapy, offering a promising tool for personalized treatment strategies.
What Problem Does This Paper Address?
Gastric cancer (GC) remains a significant global health challenge, characterized by high incidence and mortality rates, particularly in advanced stages. While current treatments like surgery, radiotherapy, and chemotherapy have improved outcomes, a substantial portion of patients still face poor prognoses. Immunotherapy has emerged as a promising avenue, but only a subset of patients benefits. Therefore, there's a critical need for accurate models that can predict GC patient prognosis, identify those likely to respond to immunotherapy, and guide treatment decisions. Traditional prognostic models often struggle with inter-dataset variability due to different sequencing methods. This paper aims to address this by developing a robust prognostic signature using immune-related long noncoding RNA pairs (IRlncRNAPs), which are hypothesized to be more stable across datasets.
Methods
The researchers employed a multi-step bioinformatics approach:
- Identification of Differentially Expressed Immune-Related Long Noncoding RNAs (DEIRlncRNAs): They analyzed RNA-seq data from The Cancer Genome Atlas (TCGA) database to identify lncRNAs that were significantly co-expressed with immune-related genes (IRGs) obtained from the Immunology Database and Analysis Portal (ImmPort).
- Construction of IRlncRNAP Pairs: They created pairs of these DEIRlncRNAs and developed a scoring system based on their relative expression levels within each sample.
- Prognostic Model Development:
- Univariate Cox Regression: Used to identify individual IRlncRNAPs significantly associated with overall survival (OS).
- LASSO Regression: Applied to select the most relevant IRlncRNAPs for building a parsimonious prognostic signature, reducing the risk of overfitting.
- Multivariate Cox Regression: Used to construct the final prognostic model and assess the independent predictive value of the identified IRlncRNAPs.
- Risk Score Calculation: A "riskScore" was calculated for each patient based on the expression levels of the selected IRlncRNAPs and their corresponding regression coefficients.
- Correlation and Validation Analyses:
- ROC Curve Analysis: Used to evaluate the sensitivity and specificity of the riskScore in predicting survival.
- Kaplan-Meier Survival Analysis: Compared survival outcomes between high-risk and low-risk groups.
- Correlation Analysis: Investigated the relationship between the riskScore and clinical features, immune cell infiltration (using data from TIMER, CIBERSORT, xCell, etc.), and expression of immune-related genes.
- Chemosensitivity Analysis: Assessed the predictive value of the riskScore for the efficacy of common chemotherapy drugs.
Datasets
- The Cancer Genome Atlas (TCGA) database: Provided RNA-seq data (FPKM) and clinical follow-up information for GC tissues and normal adjacent tissues.
- Immunology Database and Analysis Portal (ImmPort): Used to obtain a list of immune-related genes (IRGs).
- Tumor Immune Estimation Resource (TIMER), CIBERSORT, xCell, MCP-counter, and EPIC: Used for immune infiltrate estimation data.
The study did not introduce new benchmark datasets but rather utilized publicly available resources.
Key Insights and Ideas
- IRlncRNAPs as robust biomarkers: The study highlights that using pairs of immune-related lncRNAs, rather than individual ones, can create a more stable and predictive prognostic signature, potentially overcoming limitations of single-gene models.
- Immune microenvironment linkage: The riskScore derived from the IRlncRNAP signature is strongly correlated with the tumor immune microenvironment (TIME), showing associations with both pro-tumor and anti-tumor immune cell infiltration. This suggests the model captures crucial aspects of the tumor's immune landscape.
- Predictive power beyond survival: The model not only predicts survival but also offers insights into immune cell infiltration patterns and potential chemosensitivity, making it a more comprehensive tool for clinical decision-making.
- Independent prognostic factor: The IRlncRNAP signature was validated as an independent prognostic factor, meaning it provides valuable predictive information even when considering other clinical variables.
Key Results
- Identification of DEIRlncRNAs: 107 DEIRlncRNAs were identified, forming 4,297 IRlncRNAP pairs.
- Prognostic Model: A signature consisting of 15 OS-related IRlncRNAPs was developed.
- Predictive Accuracy: The 1-, 2-, 3-, and 5-year AUC values for the riskScore were 0.788, 0.810, 0.825, and 0.868, respectively, indicating high sensitivity and accuracy.
- Risk Stratification: Patients were successfully stratified into high- and low-risk groups, with the high-risk group exhibiting significantly poorer overall survival (P < 0.001).
- Correlation with Clinical Features: The riskScore was significantly correlated with tumor stage, M stage, T stage, and tumor differentiation grade, with higher scores associated with more advanced disease.
- Immune Cell Infiltration: The riskScore positively correlated with pro-tumor immune cells (e.g., CAFs, M2 macrophages) and negatively with anti-tumor immune cells (e.g., CD4+ T cells, NK cells).
- Chemosensitivity: Patients in the high-risk group showed potentially better sensitivity to docetaxel and rapamycin (lower IC50) and lower sensitivity to mitomycin C (higher IC50), suggesting potential guidance for drug selection.
Limitations and Future Work
- External Validation: The authors acknowledge that the model's reliability needs further validation using external datasets. They note the scarcity of lncRNA data in public repositories like GEO and ICGC for GC, making immediate external validation challenging.
- Functional Verification: The specific functions and mechanisms of the identified IRlncRNAs require experimental verification through in vitro and in vivo studies.
- Retrospective Nature: The study is retrospective, and prospective clinical studies are recommended to confirm the findings and clinical utility.
- Future Directions: The authors hope their model can be applied in clinical settings to guide immunotherapy and chemotherapy management and improve survival prediction for GC patients. They also suggest future large-scale sequencing studies on their own patient cohorts.
Summary updated