Single-Cell RNA Sequencing and Machine Learning Reveal Relationship Between CD8+ T Cells and Uveal Melanoma Metastasis

Oncology Medicine
single-cell RNA sequencing uveal melanoma metastasis risk immune microenvironment CD8+ T cells machine learning tumor microenvironment

Academic Report on “Machine Learning and Single-cell RNA Sequencing Reveal Relationship Between Intratumor CD8+ T Cells and Uveal Melanoma Metastasis” published in Cancer Cell International in 2024

Research Background and Purpose

Uveal melanoma (UM) is the most common intraocular malignant tumor in adults. After receiving radiation or surgical treatment, the local recurrence rate of primary UM is low. However, about 40% of patients develop distant metastasis, especially in the liver, following treatment, leading to a mortality rate of up to 50% within 4 to 5 years. Currently, the assessment of UM metastasis risk mainly relies on multigene models using machine learning methods, which predict the risk of metastasis and prognosis based on gene expression profiling (GEP) data of patients. However, these models require a large number of genes, resulting in high detection costs and difficulties in large-scale clinical applications. Furthermore, these models provide limited explanations for specific metastasis mechanisms, especially regarding how the uveal immune microenvironment affects tumor metastasis, which remains unclear. Therefore, there is an urgent need to develop a model with fewer genes that is clinically feasible while deeply exploring the mechanisms of UM metastasis.

The core aim of this study is to construct a prognostic model consisting of a small number of genes to assess the metastasis risk of UM patients. To this end, the research team combined machine learning algorithms and single-cell RNA sequencing data to explore the immunomicroenvironmental mechanisms influencing UM metastasis from the cellular type and functional levels. The research results will provide theoretical support for clinical risk assessment in UM patients and serve as a reference for the development of UM immunotherapy.

Research Source

This study was jointly completed by Shuming Chen, Zichun Tang, and others from the Department of Ophthalmology at Second Xiangya Hospital, Hunan Province, China, and was published in Cancer Cell International in 2024. Shuming Chen and Zichun Tang are co-first authors, and the corresponding authors are Xiao Liu and Zhuo Li.

Research Methods and Process

1. Data Collection and Processing

The study utilized RNA-seq data and clinical information of 79 UM patients from the TCGA database to construct and validate the prognostic model. Additionally, two sets of single-cell data (GSE138665 and GSE139829) from the GEO database were used to explore the specific mechanisms of UM metastasis. Patients were divided into metastatic and non-metastatic groups based on metastasis status, using R packages like “DESeq2,” “EdgeR,” and “Limma” to screen 247 genes associated with metastasis. Subsequent functional enrichment analysis revealed that these genes primarily participated in pathways related to the extracellular matrix microenvironment, such as the IL-17 pathway and metabolic pathways.

2. Establishment of the Prognostic Model

Through Log-rank tests and univariate Cox regression, 117 genes associated with prognosis were screened, and then a three-gene model consisting of SLC25A38, EDNRB, and LURAP1 genes was constructed using Lasso regression and multivariate Cox regression. These genes were protective factors for UM patients and demonstrated high predictive accuracy in Kaplan-Meier survival analysis. ROC curve validation showed that the model had good stability and accuracy in predictions for 6, 18, and 30 months.

3. Validation via Cellular Experiments

At the cellular level, the research team conducted cell scratch assays and CCK-8 cell viability tests, confirming that the expression of SLC25A38, EDNRB, and LURAP1 genes significantly inhibited the migratory ability of UM cells but had little effect on cell proliferation and apoptosis. These results indicate that the diagnostic biomarkers of the three-gene model play an important regulatory role in the migration and metastasis risk of UM cells.

4. Immune Microenvironment Analysis

Using R packages such as ESTIMATE and CIBERSORT, the study evaluated the immune infiltration of the groups, finding a significant increase in CD8+ T cell infiltration levels in the high metastasis risk group, but most CD8+ T cells were exhausted and functionally weak. This suggests that although the number of CD8+ T cells increased in the high-risk group, the proportion of functional CD8+ T cells was low, possibly contributing to their high risk of metastasis.

5. Single-cell RNA Sequencing Reveals Intercellular Communication

Single-cell transcriptome analysis further revealed the intercellular communication in the UM tumor microenvironment. CellChat analysis showed a significant increase in both the number and intensity of cellular communications in the high-risk group, particularly with enhanced communication intensity of cytotoxic CD8+ T cells in tumor cells. Further pathway analysis identified important pathways like APP, MHC-I, CD99, and MIF, speculating that CD99 may influence communication between CD8+ T cells and tumor cells through signaling, thereby driving UM metastasis.

6. Differentiation Trajectory of CD8+ T Cells

Through pseudotime analysis and differentiation marker gene analysis, the study found that CD8+ T cells in the high-risk group were more likely to be in the early differentiation stage with weaker functionality, demonstrating higher exhaustion status. The functional deficiency of CD8+ T cells in the high-risk group led to reduced antitumor ability, possibly being a key factor in UM metastasis.

Research Results and Main Findings

  1. Construction of Three-gene Prognostic Model: The study developed a concise model consisting of SLC25A38, EDNRB, and LURAP1 genes to assess the metastasis risk of UM patients. The model demonstrated high prediction accuracy and stability in both internal and external validation datasets, with high clinical applicability.

  2. Key Role of CD8+ T Cells: CD8+ T cells in the high-risk group exhibited significant exhaustion status, with their role in intercellular communication significantly enhanced, and the potential role of the CD99 signaling pathway in UM metastasis was confirmed.

  3. Heterogeneity of the Immune Microenvironment: The UM metastasis process is closely related to its unique immune microenvironment. CD8+ T cells in the high-risk group increased but were functionally weak, highlighting the complexity of UM immune evasion mechanisms. The study found that signaling pathways such as MHC-I and APP might be involved in antigen presentation and cell migration regulation of UM tumor cells.

  4. Impact of CD8+ T Cell Differentiation and Exhaustion Status: The increased exhaustion status of CD8+ T cells in the high-risk group played an important role in UM immune evasion. Future methods blocking CD8+ T cell exhaustion could develop more effective UM immunotherapy methods.

Research Significance

The three-gene model developed in this study provides a reliable tool for metastasis risk assessment and prognosis prediction in UM patients. Its feature of low gene count reduces gene detection costs and enhances the model’s clinical applicability. Additionally, the study deeply explored the UM immune microenvironment for the first time at the single-cell level, revealing the critical role of CD8+ T cells in the UM metastasis process and providing a new research direction for UM immunotherapy. Future strategies can design more precise immunotherapy strategies by regulating the exhaustion status of CD8+ T cells in the immune microenvironment to improve the survival rate of UM patients.