MASP1 as a favorable prognostic biomarker in pediatric osteosarcoma: an integrated analysis of machine learning, bioinformatics, and validation experiments.

Abstract

Background: Osteosarcoma (OS), the most common pediatric bone tumor, faces challenges with frequent relapse despite treatment advances. Identifying early diagnostic biomarkers and therapeutic targets is critical. The purpose of this study was to investigate the novel biomarkers for OS, and we also aimed to explore whether these biomarkers could potentially serve as the therapy targets.

Methods: Integrated analysis combined three Gene Expression Omnibus (GEO) datasets (GSE42352, GSE126209, GSE12865) and TARGET-OS clinical-transcriptomic data (n=88). Immune-related genes from ImmPort (1,793 genes) were analyzed alongside differentially expressed genes (DEGs) identified via sva batch correction. Functional enrichment used clusterProfiler, while machine learning [eXtreme Gradient Boosting (XGB), random forest (RF), generalized linear model (GLM), support vector machine (SVM)] models were built with caret, xgboost, and kernlab. Prognostic genes were screened via univariate Cox regression (P<0.05). Key genes intersecting SVM and Cox results were validated via package for receiver operating characteristic (pROC), survival analysis, competing endogenous RNA (ceRNA) network (Cytoscape), immune infiltration (CIBERSORT), drug sensitivity (GDSC), and quantitative polymerase chain reaction (qPCR).

Results: Differential analysis identified 1,370 DEGs (748 upregulated, 622 downregulated), intersecting with immune-related genes to yield 174 OS-linked immune-DEGs. Enrichment highlighted cytokine-PI3K-Akt pathways. Machine learning prioritized 10 genes, with MASP1 showing highest diagnostic accuracy [area under the curve (AUC) =0.903, 95% confidence interval (CI): 0.769-0.993]. Univariate Cox linked NRP3, STC2, ANGPT1, MASP1, SDC4, NEDD4, TYROBP to prognosis (P<0.05). Intersection identified MASP1 as the core gene, significantly downregulated in OS tissue. Survival analysis across GEO/TARGET confirmed higher MASP1 expression correlated with better outcomes (P<0.05). MASP1 inversely correlated with resting CD4+ T-cell infiltration (r=-0.14, P=0.04), a poor prognostic marker. Drug sensitivity analysis associated MASP1 with enhanced response to doxorubicin, vinblastine, gemcitabine, and sorafenib. qPCR validated MASP1 downregulation in OS samples.

Conclusions: MASP1 is a promising diagnostic biomarker and therapeutic target for OS. These findings could help to improve patient prognosis and the treatment response. Further studies should be conducted explore MASP1 clinical applications.