Prognostic significance of senescence-related tumor microenvironment genes in head and neck squamous cell carcinoma

Young Chan Lee, Yonghyun Nam, Minjeong Kim, Su Il Kim, Jung‐Woo Lee, Young‐Gyu Eun, Dokyoon KimAging 2023 · DOI

TL;DR

This study identifies a set of "senescence-related tumor microenvironment genes" (STCGs) that are crucial for predicting head and neck squamous cell carcinoma (HNSC) prognosis and response to immunotherapy. By developing a novel risk model based on these genes, researchers can stratify patients into high-risk and low-risk groups, offering potential for more personalized treatment strategies.

What Problem Does This Paper Address?

The tumor microenvironment (TME) plays a critical role in cancer progression and treatment response, but its complex interplay with cellular senescence, a state of irreversible cell cycle arrest, remains poorly understood, especially in head and neck squamous cell carcinoma (HNSC). Specifically, it's unclear whether genes associated with senescence within the TME can serve as reliable biomarkers for predicting patient outcomes and their likelihood of responding to immunotherapy. This knowledge gap hinders the development of targeted therapies and personalized treatment plans for HNSC patients.

Methods

The researchers employed a multi-pronged approach:

  • Gene Identification: They integrated data from tumor-associated senescence (TAS) genes, TME-related genes, and immune-related genes to identify a comprehensive list of prognostic senescence-related TME genes (PSTGs). This involved network analysis and label propagation to uncover interconnected gene lists.
  • Subtype Discovery: Consensus clustering was used on the expression levels of PSTGs to classify HNSC patients into distinct subtypes. This unsupervised method helps identify natural groupings within the data.
  • Risk Model Development: A Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression model was built to select a core set of senescence-related TME core genes (STCGs) for a prognostic risk score. This technique is effective for feature selection and building predictive models.
  • Validation: The developed risk model was validated using independent cohorts to ensure its robustness and generalizability.
  • Single-Cell Analysis: Single-cell RNA sequencing data was used to pinpoint the specific cell types within the TME where STCGs are expressed, providing insights into their cellular origins and functions.
  • Immune Profiling: Tools like CIBERSORT and ESTIMATE were used to analyze immune cell infiltration and TME composition, while TIDE analysis predicted immunotherapy response.

Datasets

  • TCGA-HNSC: The Cancer Genome Atlas Head and Neck Squamous Cell Carcinoma dataset was the primary source for gene expression, somatic mutation, methylation, and clinical data.
  • GEO (GSE41613): A microarray dataset from the Gene Expression Omnibus was used as an independent validation cohort.
  • KHUMC cohort: A local cohort of HNSC patients from Kyung Hee University Medical Center was used for validation.
  • TISCH database: Used for single-cell RNA sequencing data analysis of HNSC samples.

Key Insights and Ideas

  • Senescence as a Double-Edged Sword: The study reinforces the complex role of senescence, suggesting that while it can be a defense mechanism, its dysregulation within the TME can promote cancer progression and immune evasion.
  • Subtype Dichotomy: The identification of two distinct HNSC subtypes based on PSTG expression highlights a fundamental difference in tumor biology, influencing prognosis, genetic alterations, and immune infiltration.
  • Stromal and Immune Cell Dominance: STCGs are not primarily expressed in cancer cells but are enriched in stromal cells (like fibroblasts) and immune cells, suggesting that the tumor's microenvironment, rather than the tumor cells themselves, drives the senescence-related phenotype.
  • Predictive Power of the Risk Model: The developed STCG-based risk score effectively stratifies patients into high- and low-risk groups, demonstrating significant prognostic value and potential for predicting immunotherapy response.

Key Results

  • 91 Prognostic Genes Identified: A set of 91 prognostic senescence-related TME genes (PSTGs) were identified.
  • Two Distinct Subtypes: Consensus clustering revealed two subtypes (Subtype 1 and Subtype 2) with significant differences in clinical and molecular characteristics, including HPV status, tumor stage, and mutation profiles. Subtype 1 was generally associated with poorer prognosis.
  • Risk Model Validation: The STCG-based risk model successfully stratified patients into high- and low-risk groups, with the high-risk group exhibiting significantly poorer overall survival in the TCGA test set (p=0.026) and the GEO cohort (p=0.049).
  • Immunotherapy Response Prediction: The low-risk group showed a trend towards better immunotherapy response, with a higher proportion of responders predicted.
  • STCGs in Stromal and Immune Cells: Single-cell analysis revealed that risk-associated STCGs were highly expressed in fibroblasts and endothelial cells, while protective STCGs were enriched in T cells.

Limitations and Future Work

  • Protein Product Validation: The study acknowledges that protein products of the risk model genes were not validated, suggesting future research should include immunohistochemical testing or western blots.
  • Clinical Parameter Integration: While the STCG risk model shows promise, its diagnostic performance could be improved by integrating relevant clinical parameters.
  • Prospective Validation: The authors emphasize the need for prospective validation of the risk models using multicenter studies to enhance their applicability.
  • Mechanistic Investigations: Further comprehensive investigations are required to fully unravel the specific mechanisms underlying the actions of senescence-associated genes within the TME.
  • HPV Status Disparity: While the model showed efficacy in HPV-negative patients, the initial observed disparity in HPV status between subtypes warrants further exploration.

Summary updated

Abstract

The impact of the senescence related microenvironment on cancer prognosis and therapeutic response remains poorly understood. In this study, we investigated the prognostic significance of senescence related tumor microenvironment genes (PSTGs) and their potential implications for immunotherapy response. Using the Cancer Genome Atlas- head and neck squamous cell carcinoma (HNSC) data, we identified two subtypes based on the expression of PSTGs, acquired from tumor-associated senescence genes, tumor microenvironment (TME)-related genes, and immune-related genes, using consensus clustering. Using the LASSO, we constructed a risk model consisting of senescence related TME core genes (STCGs). The two subtypes exhibited significant differences in prognosis, genetic alterations, methylation patterns, and enriched pathways, and immune infiltration. Our risk model stratified patients into high-risk and low-risk groups and validated in independent cohorts. The high-risk group showed poorer prognosis and immune inactivation, suggesting reduced responsiveness to immunotherapy. Additionally, we observed a significant enrichment of STCGs in stromal cells using single-cell RNA transcriptome data. Our findings highlight the importance of the senescence related TME in HNSC prognosis and response to immunotherapy. This study contributes to a deeper understanding of the complex interplay between senescence and the TME, with potential implications for precision medicine and personalized treatment approaches in HNSC.
BibTeX
@inproceedings{lee2023prognostic,
  author = {Young Chan Lee and Yonghyun Nam and Minjeong Kim and Su Il Kim and Jung‐Woo Lee and Young‐Gyu Eun and Dokyoon Kim},
  title = {{Prognostic significance of senescence-related tumor microenvironment genes in head and neck squamous cell carcinoma}},
  booktitle = {Aging},
  year = {2023},
  doi = {10.18632/aging.205346},
}
Figure