Scholarlink

Background <p>Bladder cancer ranks among the top four most common malignancies in men worldwide. Despite therapeutic advancements, metastatic cases remain associated with dismal survival rates, highlighting the critical importance of studying environmental carcinogenic factors such as 4-aminobiphenyl (4-ABP). Classified as a Group 1 carcinogen by IARC, this compound—found in tobacco smoke and industrial chemicals—induces DNA damage through aryl amine metabolism. However, its cell-type-specific molecular mechanisms in bladder carcinogenesis remain poorly characterized, particularly at single-cell resolution.</p> Methods <p>Our study combined single-cell RNA sequencing data (<i>n</i> = 95,136 cells) with publicly available bulk transcriptomic datasets from the GEO repository. The computational workflow incorporated Harmony algorithm for batch effect correction, WGCNA for co-expression network construction, and a machine learning framework utilizing LASSO, SVM, and Random Forest algorithms for biomarker identification. Additionally, we performed molecular docking simulations to investigate 4-ABP-protein interactions and employed ssGSEA to characterize immune cell infiltration patterns in the tumor microenvironment.</p> Results <p>Single-cell profiling uncovered distinct fibroblast and mast cell subpopulations exhibiting significant 4-ABP-associated transcriptional signatures (adjusted <i>p</i> = 2.22 × 10<sup>−15</sup>). Cross-platform integration revealed 15 functionally conserved genes significantly enriched in IL-4-mediated signaling pathways (false discovery rate &lt; 5%). Through machine learning-based feature selection, we established a diagnostic panel comprising six core genes (EIF4G2, CA2, CDKN2A, HSP90B1, GOT2, and IL4R), with the random forest classifier demonstrating optimal discriminatory performance (area under curve = 1.00). Structural modeling predicted high-affinity interactions between 4-ABP and both HSP90B1 (binding energy: − 6.7&#xa0;kcal/mol) and IL4R (− 6.0&#xa0;kcal/mol). Unsupervised clustering delineated two clinically relevant molecular subtypes characterized by divergent immune microenvironment compositions.</p> Conclusion <p>Through integrated multi-omics analyses, this investigation systematically elucidates 4-ABP’s carcinogenic mechanisms while identifying clinically actionable biomarkers and molecular targets for precision medicine applications in bladder cancer prevention and treatment. These findings provide critical insights for developing targeted strategies to mitigate environmental carcinogenesis in susceptible populations. However, this study is limited by its retrospective nature and reliance on public sequencing data, which restricted access to detailed clinical metadata and precluded survival analysis for the identified subtypes.</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 single-cell transcriptomics and machine learning reveals 4-aminobiphenyl exposure signatures and novel diagnostic biomarkers in bladder cancer

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月4日

查看全文
  • Yangyang Xu,
  • Xiujia Wang,
  • Xinxin Wang,
  • Kaiqiang Li,
  • Xiuwang Wei

摘要

Background

Bladder cancer ranks among the top four most common malignancies in men worldwide. Despite therapeutic advancements, metastatic cases remain associated with dismal survival rates, highlighting the critical importance of studying environmental carcinogenic factors such as 4-aminobiphenyl (4-ABP). Classified as a Group 1 carcinogen by IARC, this compound—found in tobacco smoke and industrial chemicals—induces DNA damage through aryl amine metabolism. However, its cell-type-specific molecular mechanisms in bladder carcinogenesis remain poorly characterized, particularly at single-cell resolution.

Methods

Our study combined single-cell RNA sequencing data (n = 95,136 cells) with publicly available bulk transcriptomic datasets from the GEO repository. The computational workflow incorporated Harmony algorithm for batch effect correction, WGCNA for co-expression network construction, and a machine learning framework utilizing LASSO, SVM, and Random Forest algorithms for biomarker identification. Additionally, we performed molecular docking simulations to investigate 4-ABP-protein interactions and employed ssGSEA to characterize immune cell infiltration patterns in the tumor microenvironment.

Results

Single-cell profiling uncovered distinct fibroblast and mast cell subpopulations exhibiting significant 4-ABP-associated transcriptional signatures (adjusted p = 2.22 × 10−15). Cross-platform integration revealed 15 functionally conserved genes significantly enriched in IL-4-mediated signaling pathways (false discovery rate < 5%). Through machine learning-based feature selection, we established a diagnostic panel comprising six core genes (EIF4G2, CA2, CDKN2A, HSP90B1, GOT2, and IL4R), with the random forest classifier demonstrating optimal discriminatory performance (area under curve = 1.00). Structural modeling predicted high-affinity interactions between 4-ABP and both HSP90B1 (binding energy: − 6.7 kcal/mol) and IL4R (− 6.0 kcal/mol). Unsupervised clustering delineated two clinically relevant molecular subtypes characterized by divergent immune microenvironment compositions.

Conclusion

Through integrated multi-omics analyses, this investigation systematically elucidates 4-ABP’s carcinogenic mechanisms while identifying clinically actionable biomarkers and molecular targets for precision medicine applications in bladder cancer prevention and treatment. These findings provide critical insights for developing targeted strategies to mitigate environmental carcinogenesis in susceptible populations. However, this study is limited by its retrospective nature and reliance on public sequencing data, which restricted access to detailed clinical metadata and precluded survival analysis for the identified subtypes.