Identification and validation of neural infiltration-associated biomarkers and their potential regulatory mechanisms in osteosarcoma

Osteosarcoma (OS) is a highly aggressive bone malignancy with a poor prognosis, where the tumor microenvironment plays a critical role in tumor progression. However, the molecular biomarkers related to neuroinfiltration and their potential functions in OS remain inadequately explored. This study aimed to identify neural infiltration-associated biomarkers in OS and assess their biological functions, immune interactions, and therapeutic implications. Transcriptomic data from GEO datasets GSE99671 (training set) and GSE87624 (validation set) were analyzed to identify differentially expressed neural-related genes (NRDEGs) using DESeq2. Neural-lineage transcriptomic features scores were calculated via GSVA, and protein–protein interaction networks were constructed using STRING and Cytoscape. Core biomarkers were identified through machine learning approaches, including Boruta, random forest, and SVM–RFE algorithms. Prognostic value was assessed using Kaplan–Meier and Cox regression analyses in the GSE21257 dataset. Functional enrichment (GO, KEGG, GSEA) and immune infiltration analyses (CIBERSORT and quanTIseq) were performed. The expression and functional roles of LEPR and DLGAP1 were further validated in vitro using RT-qPCR, Western blot, CCK-8, colony formation, and Transwell assays. A total of 48 NRDEGs were identified. The C-Neural score significantly distinguished OS from normal tissues, demonstrating strong discriminatory power. Integrative machine learning identified CDH1, CXCL12, DLGAP1, and LEPR as feature genes, with DLGAP1 and LEPR consistently downregulated in validation sets. Survival analysis showed that high LEPR expression was significantly associated with improved overall survival (P = 0.011). The nomogram model integrating LEPR expression, age and gender exhibited favorable predictive performance for osteosarcoma metastasis with an AUC of 0.774, outperforming the predictive efficacy of LEPR expression alone (AUC = 0.689). GSEA revealed that LEPR was significantly enriched in the neurotrophin signaling pathway. In vitro experiments confirmed that overexpression of LEPR or DLGAP1 significantly inhibited OS cell proliferation, colony formation, migration, and invasion. Immune analysis consistently showed that LEPR was significantly negatively correlated with M2 and M1 macrophage infiltration. Drug sensitivity analysis indicated that LEPR expression was positively correlated with sensitivity to targeted agents such as SHP-099 and Lenvatinib. DLGAP1 and LEPR are key biomarkers associated with neural-lineage transcriptomic features in OS. Both act as tumor suppressors, inhibiting malignancy and modulating the immune microenvironment. LEPR further serves as a robust prognostic indicator. This study provides novel insights into the neuro-immune regulation of OS and suggests potential therapeutic targets for personalized treatment.