Page 77 - TD-4-2

Tumor Discovery DRGs in HCC prognosis and immunity

categorized into high-risk (≥mean) and low-risk (<mean) 1-, 3-, and 5-year OS in patients with HCC. Consistency
groups based on these mean risk scores. Survival curves indices and a calibration curve were employed to assess the
for these subgroups were then plotted. The “survminer” predictive accuracy of the nomogram.
package in R software generated Kaplan-Meier curves,
risk line plots, and scatterplots. These visualizations were 2.5. Functional enrichment analyses and
used to compare risk score variations among patient protein‑protein interaction
subtypes with DRGs and to illustrate the relationship Gene ontology (GO) and Kyoto Encyclopedia of Genes
between risk levels and patient survival status. Receiver and Genomes (KEGG) enrichment analyses were
operating characteristic (ROC) curve analysis, including employed to investigate differences in gene functions and
survival curves at 1, 3, and 5 years, was performed using pathways between subgroups identified by the risk model.
the “survminer” and “timeROC” packages in R software to Eighteen DRGs were identified between the low- and high-
assess the prognostic power of the risk model. In addition, risk groups in the TCGA dataset. To better understand
risk scores and clinical factors were integrated to construct the biological pathways and functions of the identified
univariate and multivariate Cox regression models, which differentially expressed genes, GO enrichment analysis
were used to evaluate the independent prognostic value of (molecular function [MF], biological process [BP], and
risk scores for HCC patients. cellular component [CC]) and KEGG pathway analysis
were conducted using the “ClusterProfiler,” “org.Hs.eg.db,”
2.3. Analysis of the correlation between prognostic “enrichplot,” and “ggplot2” packages of R software. A p<0.05
and clinical features was considered indicative of significant enrichment.
Risk scores and clinical data were integrated and analyzed Protein-protein interactions (PPIs) were visualized by
using the chi-square test to examine the correlation between submitting differentially expressed DRGs to the STRING
prognostic and clinical features. The chi-square test was database (http://www.string-db.org/). The most significant
employed to investigate the involvement of prognostic modules (the top ten highest-rated genes) within the PPI
characteristics in the development of HCC. Heatmaps network were selected using the maximal clique centrality
displaying DRGs with clinic-specific data were generated. Box algorithm through the “CytoHubba” plugin in the
plots based on risk scores were created to analyze the variance Cytoscape software (version 3.8.0).
in clinical factors and the correlation between prognostic and
clinical characteristics. Variables such as age, sex, pathologic 2.6. GSEA
N, pathologic M, pathological stage, and pathologic T To investigate the potential molecular mechanisms
were considered to determine statistical significance across distinguishing the low-risk and high-risk groups, GSEA
subgroups. Pathologic T, N, and M refer to tumor, node, was performed using the “ClusterProfiler,” “enrichplot,”
and metastasis, respectively. In addition, further stratified “ggplot2,” and “org.Hs.eg.db” packages of the R software.
analyses were performed to assess the prognostic significance A p<0.05 was considered indicative of significant
of DRG characteristics within subgroups categorized by age enrichment. Pathways with the top five highest numbers of
(≤65 years vs. >65 years), sex (male vs. female), pathologic molecules in the gene set were selected for mapping.
stage (stages I – II vs. stages III – IV), tumor grade (low vs.
high), pathologic T (stages T1 – T2 vs. stages T3 – T4), and 2.7. The relationship between prognostic signatures
pathologic N (N0 vs. N1–2–3). and immune checkpoints
Given the significance of ICI immunotherapy, the
2.4. Nomogram establishment based on risk score association between two subgroups of HCC patients,
and clinical variables
categorized by risk scores and 79 immune checkpoints, was
Univariate and multivariate Cox regression analyses and investigated, and the differences in immune checkpoint
additional clinical variables were incorporated to assess expression between these subgroups were assessed. Gene
whether risk scores possess independent prognostic value expression associated with immune checkpoints was
in predicting outcomes. Column-line plots were generated analyzed using the “limma” and “ggpubr” packages in the
to evaluate the predicted likelihood of overall survival R software, focusing on the differences between the high-
(OS). Clinical variables and risk scores were collected, risk and low-risk groups.
and Cox regression modeling was applied to calculate
hazard ratios (HRs) for each variable. The relationships 2.8. Correlation analysis between genes and
between DRG-based and clinical variables were examined. immune infiltration
Column-line plots were constructed using clinical factors Immune infiltration refers to the distribution and activity
and DRG-based risk ratings to evaluate the likelihood of of immune cells and related molecules within tumor

Volume 4 Issue 2 (2025) 69 doi: 10.36922/td.8214