Page 98 - EJMO-9-3

Eurasian Journal of
Medicine and Oncology Novel senescence-based melanoma risk model

Where n indicates the number of selected genes, exp i highlight the potential utility of senescence-related
indicates the normalized expression level of gene i in each gene expression in refining patient stratification beyond
sample, and coef indicates the corresponding coefficient conventional staging systems. Interestingly, several
i
of gene i derived from the multivariate Cox regression prognostic-related genes, known for their pivotal roles in
model. This risk score served as a quantitative measure of activating both innate and adaptive immune responses,
each patient’s prognostic risk. Based on the median risk were significantly upregulated in cluster 2 compared to
score, the training set samples were stratified into high-risk cluster 1 (Figure 1D). Notably, C-C motif chemokine
and low-risk groups. KM survival curves were generated, ligands 2 (CCL2), C-C motif chemokine ligands 4 (CCL4),
and the log-rank test, a non-parametric method for C-C motif chemokine ligands 5 (CCL5), and CXC motif
comparing survival distributions, was conducted to assess chemokine ligand 10 (CXCL10) serve as key chemokines
the survival outcomes between these groups. To assess that bind to their respective receptors, mediating a wide
the generalizability of the risk model, risk scores were range of immune processes, including the recruitment of
calculated for the validation dataset and two independent monocytes and T lymphocytes to the TME. Their elevated
external datasets (GSE65904 and GSE19234) obtained from expression in Cluster 2 suggests a more immunoreactive
the GEO database. Patients in these external cohorts were phenotype, which may contribute to improved survival
similarly categorized into high-risk and low-risk groups outcomes in this subgroup. However, it is important to
depending on the median risk score. Survival differences note that these chemokines exhibit complex and context-
were assessed using KM curves and the log-rank test, with dependent roles in tumor progression. While their
DSS as the primary endpoint in the GSE65904 dataset. involvement in immune cell recruitment and activation
supports anti-tumor immunity, their dysregulation has
3. Results also been linked to tumor-promoting mechanisms, such as
3.1. Classification of patient subtypes based on immune evasion and chronic inflammation. 29,30 Therefore,
prognostic senescence-related genes careful consideration is required when evaluating their
Processed SKCM RNA-seq data were obtained from UCSC potential clinical applications as therapeutic targets.
Xena, and 413 samples were retained after excluding Overall, these findings provide new insights into the
those without complete clinical information, including potential anti-tumor function of cellular senescence,
age, gender, stage, OS time, and other relevant variables. suggesting that senescence-related gene expression may
Given the close relationship between cellular senescence not only influence tumor progression but also modulate
and SKCM progression, 780 senescence-related genes the immune landscape of melanoma. Future research
were curated from previous studies 25-27 and assessed for is needed to clarify the mechanistic interplay between
prognostic relevance. All samples were randomly divided cellular senescence and immune activation in the TME.
into a training set and a validation set at a 7:3 ratio. Through 3.2. Immunoactivated subtypes are associated with
univariate and subsequent multivariate Cox regression prolonged survival
analysis, 190 senescence-related genes were identified as
significantly prognostic in the training set. Interestingly, To explore the functional characteristics of these clusters,
consensus clustering analysis was performed based on the pathway enrichment analysis was conducted on differentially
expression profiles of these genes in the training samples. expressed genes. Most of the upregulated genes in Cluster
By evaluating the CDF and subsequently assessing the 2 were significantly enriched in immune-related pathways,
delta area values, the optimal number of stable clusters including cytokine-cytokine receptor interaction, nuclear
was determined. This classification effectively stratified factor kappa B (NF-κB) signaling, and tumor necrosis
patients into two distinct subtypes, which demonstrated factor (TNF) signaling (Figure 2A). Furthermore, several
clear differences in survival outcomes. Notably, patients cytokines, such as Fas cell surface death receptor, CXCL10,
in Cluster 2 exhibited significantly better survival rates CCL3, CCL5, and CCL2, were highly expressed in Cluster
compared to those in Cluster 1, indicating that this 2 and have been shown to serve crucial functions in
molecular classification may have potential prognostic SKCM (Figure 2B). Higher levels of CXCL10, CCL3,
implications (Figure 1A-C). A comprehensive comparison CCL5, and CCL2 have been associated with improved
of clinical characteristics, including age, tumor stage, and responses to immunotherapy and better prognosis. 31-33
T/N/M classification, revealed no significant differences To comprehensively characterize the immune landscape
between the two identified subtypes. This suggests that the across melanoma subtypes, CIBERSORT analysis was
observed differences in survival outcomes were primarily conducted using pre-identified gene signatures to estimate
driven by molecular and transcriptomic variations rather immune cell proportions in each sample. To statistically
than traditional clinical parameters. These findings validate disparities in the immune landscape between

Volume 9 Issue 3 (2025) 90 doi: 10.36922/ejmo.8574