Page 243 - EJMO-9-2

Eurasian Journal of
Medicine and Oncology Prognosis of colon adenocarcinoma

pathological form of colon cancer is colon adenocarcinoma were developed using weighted gene co-expression
(COAD). Oncogene activation and inactivation are two network analysis, and network topology was analyzed with
2
examples of the multistage, multigene processes involved a soft-threshold power ranging from 1 to 30. The relational
in the formation of COAD. Treatment of tumors requires matrix was then converted into the adjacent matrix and
2,3
early detection and timely aggressive excision. Patients transformed into a topological overlap matrix for average
with colorectal cancer have a 5-year survival rate of link hierarchy clustering to classify related modules.
almost 90% if they receive an early diagnosis, compared The correlation between modules and clinical traits was
to only 12% for those who have distant metastases. A assessed using Pearson methods.
4
2020 study on colorectal cancer incidence revealed a
declining trend in highly developed nations, possibly 2.3. Modeling-associated genes identification and
1,5
due to improved disease screening. Identifying suitable prognostic signature construction
biomarkers is crucial for risk assessment, early diagnosis, The LASSO regression analysis was performed on the
and treatment outcome prediction. Risk stratification in survival-related genes to avoid the collinearity of high-
COAD can be enhanced using gene expression profiles. dimensional transcriptome data to identify modeling genes.
6
However, many studies have not thoroughly examined The value of this mRNA signature for the prognosis of
the genes associated with the clinical prognosis of COAD colorectal patients was verified using the Kaplan–Meier (KM)
patients. Prognostic biomarkers can significantly influence curve and log-rank test. The modeling data set’s core genes
the risk classification of COAD patients, with high- were identified using univariate Cox regression analysis with
7
risk groups benefiting from more intensive treatment p<0.05 as the threshold. Multi-factor Cox regression analysis
to prevent undertreatment. In contrast, low-risk groups was used to analyze the risk score, which was then used to
should receive low-intensity treatment regimens to avoid classify patients into high- and low-risk groups.
overtreatment. Hence, to assess the clinical prognosis of
COAD patients, we evaluated gene modules and identified 2.4. Predictive accuracy of the risk signature
potential biomarkers. The eight-mRNA signature’s predictive significance was
The relationship between COAD and the genome was evaluated through internal and external validation.
evaluated using weighted gene co-expression network 2.5. Protein expression levels and genetic
analysis. Furthermore, the least absolute shrinkage and alterations of prognostic genes validation
8
selection operator (LASSO) Cox regression and univariate
proportional hazards analyses were used to identify an The cBioPortal database (https://www.cbioportal.org/)
mRNA signature closely linked to the prognosis of COAD, was used to investigate the mutation status of key genes
surpassing clinical criteria. Ultimately, the gene signature in the mRNA signature, whereas the Human Protein Atlas
11
and clinical characteristics were combined to create a database (https://www.proteinatlas.org/) was utilized to
nomogram to predict prognosis. verify the protein expression levels of these genes through
immunohistochemistry.
2. Materials and methods
2.6. Immune landscape differences of colon
2.1. Data retrieval process adenocarcinoma patients across risk strata
All bioinformatics and statistical analyses were performed The single-sample gene set enrichment analysis (ssGSEA)
using R software (version 4.2.0). Raw data were obtained algorithm was employed to ascertain the impact of
from the Cancer Genome Atlas Program (TCGA) (https:// high- and low-risk groups on immune cell infiltration and
portal.gdc.cancer.gov/), which included 647 colorectal immune function in the TCGA and GEO cohorts.
cancer patients with complete survival information.
From the Gene Expression Omnibus (GEO) (http://www. 2.7. Univariate and multi-factor Cox regression
ncbi.nlm.nih.gov/geo/) database, the GSE39582 dataset analysis
9
was used for validation. The TCGA dataset contained Cox regression analyses, both univariate and multivariate,
10
647 tumor samples and 51 normal samples, whereas the were conducted on age, sex, stage, TNM staging, and
GSE39582 dataset included 556 colorectal cancer samples. other clinical factors in the TCGA and GEO cohorts.
This analysis was conducted to confirm the mRNA risk
2.2. Co-expression network construction
signature’s independent predictive value. Univariate
In this study, intramodular hub genes were selected and multivariate Cox regression, KM curves, and log-
using the criterion of gene significance>0.2 and module rank tests were performed using the R package survival
membership>0.8. Weighted gene co-expression networks (version 3.3 1) and “survminer” (version 0.4.9).

Volume 9 Issue 2 (2025) 235 doi: 10.36922/EJMO025060024