Identification and preliminary validation of key exosome-related genes in nasopharyngeal carcinoma through integrative bioinformatics and experimental approaches

Abstract Exosomes have emerged as significant contributors to the progression of nasopharyngeal carcinoma (NPC). However, the characterization of exosome-related gene (ERG) signatures and their immunological relevance in NPC remains limited. This study systematically identified and validated key ERGs in NPC through integrative bioinformatics and machine-learning approaches. Datasets GSE12452 and GSE53819, in addition to known ERGs, were obtained from public databases. Identification of key ERGs employed machine learning, expression verification, and receiver operating characteristic (ROC) curve analysis. Gene set enrichment analysis (GSEA), immune infiltration analysis, in silico compound-target prediction, and molecular docking were also conducted. The expression of key genes underwent experimental validation in clinical NPC tissue samples. A total of 11 candidate genes emerged from machine learning analysis. Of these, LDHA , PIGR , and POSTN demonstrated significant differential expression and passed ROC curve evaluation in both GSE12452 and GSE53819. GSEA revealed involvement of these genes in pathways related to cell cycle checkpoints and DNA replication. Immune infiltration analysis indicated a significantly lower proportion of neutrophils in NPC samples compared to controls, alongside an increased proportion of type 2 helper T cells. Correlation analysis showed that LDHA and POSTN were negatively correlated with neutrophils, while PIGR displayed a positive correlation. Additionally, drug prediction and molecular docking (binding energy: −8.552 kcal/mol) indicated strong binding between LDHA and CHEMBL2058994. Subsequent 100-ns molecular dynamics simulations affirmed the structural and energetic stability of the LDHA –CHEMBL2058994 complex in silico, suggesting its potential as a candidate for further experimental and pharmacological evaluation in NPC. RT-qPCR validation confirmed significant differential expression for LDHA and PIGR , whereas POSTN exhibited a consistent trend without reaching statistical significance in NPC tissue. This study identified three key ERG candidate genes ( LDHA , PIGR , and POSTN ) with potential biomarker value. RT-qPCR validation confirmed significant differential expression of LDHA and PIGR , while POSTN showed a consistent upregulation trend in nasopharyngeal carcinoma tissues but did not reach statistical significance, requiring further evaluation in a larger sample cohort.