Scholarlink

Background <p>Lung adenocarcinoma (LUAD), the predominant histological subtype of non-small cell lung cancer, remains a leading cause of cancer-related mortality worldwide. The RAS signaling pathway plays a critical role in LUAD pathogenesis; however, the heterogeneity of RAS pathway activity and its clinical implications remain poorly understood. This study aimed to characterize RAS pathway activity subtypes and develop a robust prognostic model for LUAD patients.</p> Methods <p>Transcriptomic data from 624 LUAD patients (GEO datasets: GSE31210 and GSE72094) were analyzed as the training cohort, with TCGA-LUAD as validation. Consensus clustering stratified patients based on 238 RAS pathway-related genes. Candidate genes were identified through differential expression analysis and WGCNA. Machine learning algorithms (LASSO, Random Forest, SHAP) were applied to construct a prognostic risk model. Comprehensive analyses including GSEA, CIBERSORTx-based immune infiltration, ESTIMATE scoring, and drug sensitivity prediction were performed.</p> Results <p>The study identified two distinct RAS pathway activity subtypes among LUAD patients. A three-gene prognostic signature (MAPK10, PLA2G12B, SHC3) was established, with the risk score serving as an independent prognostic indicator. Risk score was an independent prognostic factor. Immune landscape analysis demonstrated that high- and low-risk patients showed different expression levels of immune cells. All signature genes correlated positively with resting mast cells, while MAPK10 and PLA2G12B negatively correlated with activated CD4 + memory T cells. GSEA revealed high-risk tumors enriched in DNA replication, cell cycle, and base excision repair, whereas low-risk tumors favored drug and retinol metabolism pathways. Low-risk patients exhibited lower TIDE and exclusion scores, indicating better immunotherapy response potential. Eight therapeutic compounds (genistein, metformin, quercetin, JNK-9&#xa0;L) demonstrated favorable binding to signature genes.</p> Conclusion <p>This study established a comprehensive landscape of RAS pathway activity subtypes in LUAD, identifying MAPK10, PLA2G12B, and SHC3 as novel prognostic biomarkers with significant associations with immune microenvironment remodeling, providing new insights for personalized treatment strategies and therapeutic target development.</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

RAS pathway activity subtypes identified by machine learning define prognostic and immune microenvironment characteristics in lung adenocarcinoma

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月28日

查看全文
  • Yao shi

摘要

Background

Lung adenocarcinoma (LUAD), the predominant histological subtype of non-small cell lung cancer, remains a leading cause of cancer-related mortality worldwide. The RAS signaling pathway plays a critical role in LUAD pathogenesis; however, the heterogeneity of RAS pathway activity and its clinical implications remain poorly understood. This study aimed to characterize RAS pathway activity subtypes and develop a robust prognostic model for LUAD patients.

Methods

Transcriptomic data from 624 LUAD patients (GEO datasets: GSE31210 and GSE72094) were analyzed as the training cohort, with TCGA-LUAD as validation. Consensus clustering stratified patients based on 238 RAS pathway-related genes. Candidate genes were identified through differential expression analysis and WGCNA. Machine learning algorithms (LASSO, Random Forest, SHAP) were applied to construct a prognostic risk model. Comprehensive analyses including GSEA, CIBERSORTx-based immune infiltration, ESTIMATE scoring, and drug sensitivity prediction were performed.

Results

The study identified two distinct RAS pathway activity subtypes among LUAD patients. A three-gene prognostic signature (MAPK10, PLA2G12B, SHC3) was established, with the risk score serving as an independent prognostic indicator. Risk score was an independent prognostic factor. Immune landscape analysis demonstrated that high- and low-risk patients showed different expression levels of immune cells. All signature genes correlated positively with resting mast cells, while MAPK10 and PLA2G12B negatively correlated with activated CD4 + memory T cells. GSEA revealed high-risk tumors enriched in DNA replication, cell cycle, and base excision repair, whereas low-risk tumors favored drug and retinol metabolism pathways. Low-risk patients exhibited lower TIDE and exclusion scores, indicating better immunotherapy response potential. Eight therapeutic compounds (genistein, metformin, quercetin, JNK-9 L) demonstrated favorable binding to signature genes.

Conclusion

This study established a comprehensive landscape of RAS pathway activity subtypes in LUAD, identifying MAPK10, PLA2G12B, and SHC3 as novel prognostic biomarkers with significant associations with immune microenvironment remodeling, providing new insights for personalized treatment strategies and therapeutic target development.