Establishment and verification of a prognostic model based on coagulation and fibrinolysis-related genes in hepatocellular carcinoma

Meng Fan, Le Lü, Hao Shang, Yuxuan Lu, Yi Yang, Xiuyan Wang, Hongwei LüAging 2024 · DOI

TL;DR

Researchers have developed a novel "RiskScore" model using genes related to blood coagulation and fibrinolysis to predict the prognosis of hepatocellular carcinoma (HCC). This model demonstrates strong predictive power, offering a potential tool for personalized treatment strategies in HCC patients.

What Problem Does This Paper Address?

Hepatocellular carcinoma (HCC) is a significant global health threat with high mortality. While treatments have advanced, many patients don't respond well to current therapies, and early diagnosis is challenging. Identifying reliable prognostic markers is crucial for better patient management and treatment optimization. Previous studies have hinted at a connection between the coagulation and fibrinolysis (CFR) systems and cancer progression, but a comprehensive model specifically for HCC prognosis based on CFR-related genes was lacking. This paper aims to fill that gap by building and validating such a model.

Methods

The study employed a multi-pronged approach to identify key genes and build a predictive model:

  • Data Acquisition and Preprocessing: RNA-Seq data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases were used. Standard preprocessing steps like log2 transformation were applied.
  • Gene Set Identification: A curated list of 149 coagulation and fibrinolysis (CFR)-correlated genes was obtained from the Molecular Signatures Database (MSigDB).
  • Differential Gene Expression Analysis: The limma package was used to identify differentially expressed genes (DEGs) between tumor and normal samples.
  • Weighted Gene Co-expression Network Analysis (WGCNA): This technique was used to identify modules of co-expressed genes and find those most strongly associated with CFR scores.
  • Key Gene Selection: A combination of univariate Cox regression, LASSO regression, and randomForest algorithms were used to narrow down the list of significant prognostic genes. Stepwise regression was then applied to select the final set of key genes.
  • RiskScore Model Development: A RiskScore was calculated for each patient using a weighted sum of the expression levels of the selected key genes, with weights derived from Cox regression coefficients.
  • Model Validation: The TCGA dataset served as the training set, and the GSE14520 dataset was used for independent validation.
  • Prognostic Evaluation: Receiver Operating Characteristic (ROC) curves and Kaplan-Meier survival curves were used to assess the model's predictive accuracy and survival prediction capabilities.
  • Clinicopathological Analysis: The relationship between RiskScore and various clinicopathological features (stage, grade, etc.) was analyzed. A nomogram was constructed to integrate RiskScore with other prognostic factors.
  • Immune Microenvironment and Drug Sensitivity Analysis: Tools like CIBERSORT and TIMER were used to assess immune cell infiltration and immune scores. TIDE analysis was employed to predict immunotherapy response, and pRRophetic was used to predict drug sensitivity.
  • In Vitro Validation: Gene expression levels of the key genes were validated in HCC cell lines (HUH7) and normal liver cells (LO-2) using qRT-PCR. The effect of PDE2A on cell migration and invasion was also tested.

Datasets

  • TCGA-HCC: Used as the primary training dataset, containing RNA-Seq data from 355 HCC samples and 52 adjacent normal tissues.
  • GSE14520: Used as an independent validation dataset, containing expression data from 221 HCC tumor samples.
  • MSigDB: Provided a curated list of 149 CFR-correlated genes.

Key Insights and Ideas

  • CFR Genes as Prognostic Indicators: The study strongly suggests that genes involved in blood coagulation and fibrinolysis are not just bystanders but active players in HCC progression and prognosis.
  • The Power of a Multi-Algorithmic Approach: Combining WGCNA with multiple regression techniques (Cox, LASSO, randomForest) proved effective in identifying a robust set of key prognostic genes.
  • RiskScore's Predictive Prowess: The developed RiskScore model demonstrated high AUC values for 1, 3, and 5-year survival predictions, indicating its potential for clinical utility.
  • Link to Immune Microenvironment: Higher RiskScores were associated with altered immune cell infiltration and expression of immune checkpoints, suggesting a connection between CFR pathways and the tumor immune landscape. This could have implications for immunotherapy response.
  • Potential for Targeted Therapies: The findings suggest that patients with high RiskScores might benefit more from traditional chemotherapy, while also hinting at potential therapeutic avenues related to the identified CFR genes.

Key Results

  • Six Key Genes Identified: The study identified six genes (PCDH17, PGF, PDE2A, FAM110D, FSCN1, and FBLN5) as crucial for predicting HCC prognosis.
  • High Predictive Accuracy: The RiskScore model achieved AUC values of 0.75 (1-year), 0.73 (3-year), and 0.69 (5-year) in the TCGA dataset, and similar performance was observed in the GSE14520 validation set.
  • Survival Correlation: A higher RiskScore was consistently associated with lower survival probability and a higher proportion of advanced clinicopathological grades.
  • Immune Correlations: High RiskScore groups showed higher immune scores and a significant correlation with immune checkpoints and immunomodulators.
  • Drug Sensitivity Differences: Patients in the low RiskScore group were more susceptible to erlotinib, while five conventional chemotherapy treatments were more advantageous for the high RiskScore group.
  • In Vitro Validation: The expression of key genes was validated in HCC cell lines, and PDE2A overexpression was shown to inhibit cell migration and invasion, supporting its role in suppressing metastasis.

Limitations and Future Work

  • Dataset Size: The authors acknowledge that the collection amount of tumor sample data and gene sample data is small, and there is a lack of complete clinicopathological studies. They suggest integrating more samples for later analysis.
  • Mechanism Unclear: The specific roles of CFR-related genes in HCC pathogenesis and their precise mechanisms are not fully elucidated, necessitating further research.
  • Broader Validation: While validated on an independent dataset, further validation in larger, diverse patient cohorts would strengthen the model's generalizability.
  • Therapeutic Strategies: Future work could focus on developing targeted therapies based on the identified CFR genes and exploring their combined effects with existing treatments. Investigating the interplay between CFR pathways and the tumor immune microenvironment in more detail is also a promising direction.

Summary updated

Abstract

We screened CFR related genes to design a RiskScore model, which could accurately evaluate the prognosis and survival status of HCC patients, providing certain value for optimizing the clinical treatment of cancer in the future.
BibTeX
@inproceedings{fan2024establishment,
  author = {Meng Fan and Le Lü and Hao Shang and Yuxuan Lu and Yi Yang and Xiuyan Wang and Hongwei Lü},
  title = {{Establishment and verification of a prognostic model based on coagulation and fibrinolysis-related genes in hepatocellular carcinoma}},
  booktitle = {Aging},
  year = {2024},
  doi = {10.18632/aging.205699},
}
Figure