Scholarlink

<p>This study aims to identify distinct sets of genes that can serve as specific biomarkers reflective of biological states distinguishing diabetic nephropathy (DN) or diabetic retinopathy (DR) from healthy individuals, thus providing potential utility for early detection and intervention in diabetic complications. We analyzed two transcriptome datasets (GSE142153 and GSE221521) from the Gene Expression Omnibus, employing the Limma method to identify differentially expressed genes (DEGs) among healthy controls, individuals with DN, and those with DR. Through gene co-regulation mechanisms, we screened for crucial genes related to both DN and DR, and selected candidate biomarkers via functional enrichment analysis, protein–protein interaction network construction, and the integration of machine learning algorithms. Their diagnostic performance was evaluated using receiver operating characteristic curves and dynamic nomogram analysis, and further validated with the GSE154881 and GSE185011 datasets. We identified 48 genes co-regulated by DM and DN, and 171 genes co-regulated by DM and DR. Using protein–protein interaction networks, the top 10 and 18 node genes for DN and DR were screened, respectively, leading to the selection of two biomarkers for DN (HCAR2 and TANK) and three for DR (RHOB, RPL12, and RPS17) through machine learning. Validation results demonstrated the good efficacy of the nomogram, suggesting that these distinct gene sets can serve as specific biomarkers for indicating distinct biological alterations associated with DN and DR.</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

Biomarker signatures distinguish diabetic retinopathy and diabetic nephropathy

  • 期刊论文

  • 发表日期：

    2025年8月6日

查看全文
  • Zhaocheng Li,
  • Yanqing Wang

摘要

This study aims to identify distinct sets of genes that can serve as specific biomarkers reflective of biological states distinguishing diabetic nephropathy (DN) or diabetic retinopathy (DR) from healthy individuals, thus providing potential utility for early detection and intervention in diabetic complications. We analyzed two transcriptome datasets (GSE142153 and GSE221521) from the Gene Expression Omnibus, employing the Limma method to identify differentially expressed genes (DEGs) among healthy controls, individuals with DN, and those with DR. Through gene co-regulation mechanisms, we screened for crucial genes related to both DN and DR, and selected candidate biomarkers via functional enrichment analysis, protein–protein interaction network construction, and the integration of machine learning algorithms. Their diagnostic performance was evaluated using receiver operating characteristic curves and dynamic nomogram analysis, and further validated with the GSE154881 and GSE185011 datasets. We identified 48 genes co-regulated by DM and DN, and 171 genes co-regulated by DM and DR. Using protein–protein interaction networks, the top 10 and 18 node genes for DN and DR were screened, respectively, leading to the selection of two biomarkers for DN (HCAR2 and TANK) and three for DR (RHOB, RPL12, and RPS17) through machine learning. Validation results demonstrated the good efficacy of the nomogram, suggesting that these distinct gene sets can serve as specific biomarkers for indicating distinct biological alterations associated with DN and DR.