Scholarlink

Background <p>Lung squamous cell carcinoma (LUSC) exhibits poor prognosis and a highly complex tumor immune microenvironment (TIME), creating an urgent clinical need for novel biomarkers to guide personalized treatment. Although kinase-related genes (KRGs) play a central role in signal transduction and malignant progression across multiple tumors, large-scale systematic exploration of KRGs’ regulatory functions in LUSC survival prognosis and microenvironmental remodeling remains scarce. This study aims to investigate the impact of KRGs on LUSC development, construct a promising predictive model, and provide theoretical support for personalized treatment strategies.</p> Methods <p>This study integrated transcriptomic and clinical data from the TCGA-LUSC cohort to identify differentially expressed and prognosis-related KRGs. Unsupervised consensus clustering identified kinase-related intrinsic subtypes in LUSC. Subsequently, LASSO-Cox regression analysis was employed to determine key KRGs and construct a risk prediction model, which was validated across two independent GEO external datasets (GSE157010 and GSE73403). Furthermore, this study combined ESTIMATE and CIBERSORT algorithms to assess immune infiltration patterns. Using single-cell RNA sequencing data (GSE127465 and GSE162498), it characterized the expression distribution patterns of key KRGs within the tumor microenvironment. By integrating gene transcriptomics data, it predicted chemotherapy drug sensitivity across different risk subgroups.</p> Results <p>The study identified 21 differentially expressed KRGs associated with prognosis, which were used to classify LUSC patients into molecular subtypes (C1-C2 and CA-CB) exhibiting significant survival differences and distinct immune landscapes. Four key KRGs (LATS2, CHEK2, TRIB1, and ROS1) were selected via machine learning, enabling the construction of a kinase-associated risk prediction model. Its area under the curve (AUC) for predicting 1-, 3-, and 5-year overall survival in the TCGA-LUSC training cohort was 0.585, 0.637, and 0.605, respectively. This predictive model was further developed and validated in two external cohorts, where the 1-year, 3-year, and 5-year AUC values hovered around 0.60, confirming its moderate and relatively stable predictive performance. Immunological analysis revealed that high-risk features were closely associated with reduced CD8 + T cell infiltration and abnormal enrichment of neutrophils and resting immune cells, collectively shaping a highly immunosuppressive microenvironment. Drug sensitivity analysis further indicated that the high-risk group exhibited stronger intrinsic resistance to multiple traditional first-line chemotherapy drugs but potentially demonstrated high sensitivity to specific agents such as topoisomerase I inhibitors.</p> Conclusion <p>A risk model comprising four key KRGs can relatively effectively predict survival outcomes in LUSC patients and accurately reflect the highly immunosuppressive and exhausted characteristics of the tumor microenvironment. This study comprehensively and systematically reveals the multidimensional synergistic network of kinase features driving malignant progression in LUSC, providing potential insights and theoretical basis for overcoming conventional chemotherapy resistance and developing personalized drug strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrating machine learning and single-cell sequencing to reveal the role of kinase-related genes in subtype classification and prognostic significance of lung squamous cell carcinoma

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月9日

查看全文
  • Yan Lv,
  • Zhihan Xiao,
  • Xu Zhan,
  • Xinji Liu,
  • Wei Tang,
  • Qihang Sun,
  • Qi Wang,
  • Ruijie Zhang,
  • Wei Ping,
  • Ni Zhang

摘要

Background

Lung squamous cell carcinoma (LUSC) exhibits poor prognosis and a highly complex tumor immune microenvironment (TIME), creating an urgent clinical need for novel biomarkers to guide personalized treatment. Although kinase-related genes (KRGs) play a central role in signal transduction and malignant progression across multiple tumors, large-scale systematic exploration of KRGs’ regulatory functions in LUSC survival prognosis and microenvironmental remodeling remains scarce. This study aims to investigate the impact of KRGs on LUSC development, construct a promising predictive model, and provide theoretical support for personalized treatment strategies.

Methods

This study integrated transcriptomic and clinical data from the TCGA-LUSC cohort to identify differentially expressed and prognosis-related KRGs. Unsupervised consensus clustering identified kinase-related intrinsic subtypes in LUSC. Subsequently, LASSO-Cox regression analysis was employed to determine key KRGs and construct a risk prediction model, which was validated across two independent GEO external datasets (GSE157010 and GSE73403). Furthermore, this study combined ESTIMATE and CIBERSORT algorithms to assess immune infiltration patterns. Using single-cell RNA sequencing data (GSE127465 and GSE162498), it characterized the expression distribution patterns of key KRGs within the tumor microenvironment. By integrating gene transcriptomics data, it predicted chemotherapy drug sensitivity across different risk subgroups.

Results

The study identified 21 differentially expressed KRGs associated with prognosis, which were used to classify LUSC patients into molecular subtypes (C1-C2 and CA-CB) exhibiting significant survival differences and distinct immune landscapes. Four key KRGs (LATS2, CHEK2, TRIB1, and ROS1) were selected via machine learning, enabling the construction of a kinase-associated risk prediction model. Its area under the curve (AUC) for predicting 1-, 3-, and 5-year overall survival in the TCGA-LUSC training cohort was 0.585, 0.637, and 0.605, respectively. This predictive model was further developed and validated in two external cohorts, where the 1-year, 3-year, and 5-year AUC values hovered around 0.60, confirming its moderate and relatively stable predictive performance. Immunological analysis revealed that high-risk features were closely associated with reduced CD8 + T cell infiltration and abnormal enrichment of neutrophils and resting immune cells, collectively shaping a highly immunosuppressive microenvironment. Drug sensitivity analysis further indicated that the high-risk group exhibited stronger intrinsic resistance to multiple traditional first-line chemotherapy drugs but potentially demonstrated high sensitivity to specific agents such as topoisomerase I inhibitors.

Conclusion

A risk model comprising four key KRGs can relatively effectively predict survival outcomes in LUSC patients and accurately reflect the highly immunosuppressive and exhausted characteristics of the tumor microenvironment. This study comprehensively and systematically reveals the multidimensional synergistic network of kinase features driving malignant progression in LUSC, providing potential insights and theoretical basis for overcoming conventional chemotherapy resistance and developing personalized drug strategies.