Scholarlink

<p>Bladder cancer is a complex disease characterized by significant age-related variations in progression and outcomes. In efforts to improve non-diagnostic approach for the disease, the focus of this study is to develop a personalized, reliable method for identifying multi-omics biomarkers that specifically account for the significant age-related variations observed in bladder cancer progression and outcomes. Current clinical practices, such as invasive urine cytology, often lack the sensitivity for low-grade tumors, while traditional computational methods for multi-omics analysis typically fail to model the underlying molecular network structure and overlook the interplay between patient age and genomic biomarkers. These limitations restrict the potential for truly personalized treatment strategies. To overcome these challenges, we propose a novel framework that utilizes a Graph Neural Network (GNN), specifically GraphSAGE, to identify age-specific multi-omics biomarkers in bladder cancer by integrating multi-omics data, including copy number alterations (CNA), DNA methylation, and mRNA expression. Our novel approach constructs graph-based representations of molecular interactions, incorporating patient age as both a stratification factor and a graph feature. This enables the model to capture age-dependent molecular signatures unique to different patient groups. Using a robust feature selection pipeline, we reduced dimensionality and identified key genomic features associated with bladder cancer progression. Our results show that GNNs, particularly GraphSAGE, outperform traditional methods, achieving an accuracy of 82.84% and an AUC of 0.8804 in predicting age-stratified survival outcomes. SHAP analysis highlights critical genes, such as SNRPN and DHX36, which contribute significantly to age-related predictions and offer new insights into the biological mechanisms underlying bladder cancer. This study advances our understanding of age-specific genomic profiles in bladder cancer, while providing a framework for personalized diagnostic and prognostic tools. Our findings have the potential to inform targeted treatment strategies for different age groups, ultimately improving patient outcomes. The proposed methodology is broadly applicable to other cancers in which age plays a key role in disease progression, paving the way for future research in precision oncology.</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

Graphsage-based approach for age-specific multi-omics biomarker identification in bladder cancer

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • Usman Fakhar,
  • Mohammad Elshafie,
  • Abedalrhman Alkhateeb

摘要

Bladder cancer is a complex disease characterized by significant age-related variations in progression and outcomes. In efforts to improve non-diagnostic approach for the disease, the focus of this study is to develop a personalized, reliable method for identifying multi-omics biomarkers that specifically account for the significant age-related variations observed in bladder cancer progression and outcomes. Current clinical practices, such as invasive urine cytology, often lack the sensitivity for low-grade tumors, while traditional computational methods for multi-omics analysis typically fail to model the underlying molecular network structure and overlook the interplay between patient age and genomic biomarkers. These limitations restrict the potential for truly personalized treatment strategies. To overcome these challenges, we propose a novel framework that utilizes a Graph Neural Network (GNN), specifically GraphSAGE, to identify age-specific multi-omics biomarkers in bladder cancer by integrating multi-omics data, including copy number alterations (CNA), DNA methylation, and mRNA expression. Our novel approach constructs graph-based representations of molecular interactions, incorporating patient age as both a stratification factor and a graph feature. This enables the model to capture age-dependent molecular signatures unique to different patient groups. Using a robust feature selection pipeline, we reduced dimensionality and identified key genomic features associated with bladder cancer progression. Our results show that GNNs, particularly GraphSAGE, outperform traditional methods, achieving an accuracy of 82.84% and an AUC of 0.8804 in predicting age-stratified survival outcomes. SHAP analysis highlights critical genes, such as SNRPN and DHX36, which contribute significantly to age-related predictions and offer new insights into the biological mechanisms underlying bladder cancer. This study advances our understanding of age-specific genomic profiles in bladder cancer, while providing a framework for personalized diagnostic and prognostic tools. Our findings have the potential to inform targeted treatment strategies for different age groups, ultimately improving patient outcomes. The proposed methodology is broadly applicable to other cancers in which age plays a key role in disease progression, paving the way for future research in precision oncology.