Scholarlink

Background <p>Lung adenocarcinoma (LUAD) represents an aggressive malignancy characterized by high metastatic potential. Emerging evidence suggests mitochondrial DNA methylation (MTDM) plays a pivotal role in regulating gene expression through protein synthesis modulation, yet its mechanistic involvement in LUAD pathogenesis remains poorly understood.</p> Methods <p>We systematically identified differentially expressed MTDM-related genes (DEMTDMRGs) through intersection analysis of differentially expressed genes and weighted gene co-expression network modules. Functional enrichment analysis was performed for the identified 339 DEMTDMRGs. Prognostic gene signatures were established using machine learning algorithms, followed by comprehensive validation of the risk model through Kaplan-Meier and ROC analyses. The clinical utility was further evaluated via nomogram construction. Immune cell infiltration patterns and drug sensitivity were analyzed across risk strata. Pathway enrichment was investigated through GSEA. Single-cell RNA sequencing elucidated cell-type specific expression patterns of prognostic genes, with subsequent experimental validation by qRT-PCR.</p> Results <p>Functional analysis revealed DEMTDMRGs were significantly enriched in cell cycle regulation, ferroptosis, and ABC transporter pathways. Our machine learning-derived prognostic model incorporating six genes (GJB3, RGS20, PTPRH, GPR37, STK32A, and CNTNAP2) demonstrated robust predictive capacity (1-year AUC = 0.82). The riskScore emerged as an independent prognostic factor (HR = 1.87, 95%CI = 1.32–2.65). Distinct immune infiltration patterns were observed between risk groups, with 15 immune cell subsets showing differential abundance. Pathway analysis identified 27 KEGG and 30 HALLMARK pathways, with particular enrichment in cell proliferation and metabolic processes. High-risk patients exhibited enhanced sensitivity to AZD7762 (<i>P</i> = 0.003). Single-cell resolution analysis localized predominant expression of five prognostic genes (GJB3, RGS20, PTPRH, GPR37, and STK32A) in epithelial cells, with elevated expression in tumor samples. Experimental validation confirmed significant overexpression of all six genes in LUAD cells.</p> Conclusion <p>Our study elucidates the multifaceted roles of MTDM in LUAD pathogenesis and establishes a novel six-gene signature with prognostic and therapeutic implications. The identified biomarkers not only predict immunotherapy response but also provide accurate risk stratification, offering new perspectives for precision oncology in LUAD management.</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

Mitochondrial DNA methylation predicts immunotherapy response and prognosis in lung adenocarcinoma: evidence from scRNA-Seq and machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月10日

查看全文
  • Jian Ding,
  • Gang Cheng,
  • Qian Xue,
  • Weizhen Guo,
  • Yikun Cheng,
  • Cheng Yang,
  • Jiabing Tong,
  • Zegeng Li,
  • Yating Gao

摘要

Background

Lung adenocarcinoma (LUAD) represents an aggressive malignancy characterized by high metastatic potential. Emerging evidence suggests mitochondrial DNA methylation (MTDM) plays a pivotal role in regulating gene expression through protein synthesis modulation, yet its mechanistic involvement in LUAD pathogenesis remains poorly understood.

Methods

We systematically identified differentially expressed MTDM-related genes (DEMTDMRGs) through intersection analysis of differentially expressed genes and weighted gene co-expression network modules. Functional enrichment analysis was performed for the identified 339 DEMTDMRGs. Prognostic gene signatures were established using machine learning algorithms, followed by comprehensive validation of the risk model through Kaplan-Meier and ROC analyses. The clinical utility was further evaluated via nomogram construction. Immune cell infiltration patterns and drug sensitivity were analyzed across risk strata. Pathway enrichment was investigated through GSEA. Single-cell RNA sequencing elucidated cell-type specific expression patterns of prognostic genes, with subsequent experimental validation by qRT-PCR.

Results

Functional analysis revealed DEMTDMRGs were significantly enriched in cell cycle regulation, ferroptosis, and ABC transporter pathways. Our machine learning-derived prognostic model incorporating six genes (GJB3, RGS20, PTPRH, GPR37, STK32A, and CNTNAP2) demonstrated robust predictive capacity (1-year AUC = 0.82). The riskScore emerged as an independent prognostic factor (HR = 1.87, 95%CI = 1.32–2.65). Distinct immune infiltration patterns were observed between risk groups, with 15 immune cell subsets showing differential abundance. Pathway analysis identified 27 KEGG and 30 HALLMARK pathways, with particular enrichment in cell proliferation and metabolic processes. High-risk patients exhibited enhanced sensitivity to AZD7762 (P = 0.003). Single-cell resolution analysis localized predominant expression of five prognostic genes (GJB3, RGS20, PTPRH, GPR37, and STK32A) in epithelial cells, with elevated expression in tumor samples. Experimental validation confirmed significant overexpression of all six genes in LUAD cells.

Conclusion

Our study elucidates the multifaceted roles of MTDM in LUAD pathogenesis and establishes a novel six-gene signature with prognostic and therapeutic implications. The identified biomarkers not only predict immunotherapy response but also provide accurate risk stratification, offering new perspectives for precision oncology in LUAD management.