Prognostic significance of migrasomes in neuroblastoma through machine learning and multi-omics

Abstract This study explores migrasomes' role in neuroblastoma, a common malignant tumor in children, and their potential impact on tumor formation. We analyzed neuroblastoma RNA-seq datasets from public databases, including GSE62564, GSE181559, target, and fwr144. Through data normalization and unsupervised classification using migrasome-specific molecular markers, Differentially Expressed Genes were identified, followed by functional enrichment analysis. Our novel migrasome-associated machine learning model, MigScore, was developed using ten algorithms and 101 combinations, validated on two single-cell datasets. This enabled immune infiltration assessment and drug compatibility prediction, highlighting the utility of MS275, a histone deacetylase inhibitor. Results showed a significant inverse relationship between MigScore and favorable clinical outcomes, elucidating the link between migrasome pathways and tumor immunogenicity. These findings suggest that migrasomes are crucial in neuroblastoma prognosis, leading to the possibility of personalized treatment strategies and improved outcomes.