A taurine metabolism-related gene program defines clinical and biological features of neuroblastoma and yields an 8-gene candidate prognostic model.

Taurine metabolism-related transcriptional programs have not been systematically characterized in neuroblastoma. We aimed to investigate the clinical relevance and biological features of a taurine metabolism-related gene program in neuroblastoma and to determine whether this candidate space could yield a prognostically informative gene signature. A taurine-related gene program was quantified using single-sample gene set enrichment analysis in the discovery cohort. Its associations with clinicopathologic variables, survival, and hallmark pathways were evaluated, followed by gene-level prognostic screening and construction of a compressed candidate prognostic model. Model performance was assessed in the discovery cohort and externally evaluated in independent validation datasets. Integrative prioritization was then performed to identify core candidate genes, which were further examined using Jansky-based tumor-cell pseudobulk and patient-aware phase analyses. Taurine single-sample gene set enrichment analysis (ssGSEA) score was significantly associated with Children's Oncology Group (COG) risk group, MYCN status, International Neuroblastoma Staging System (INSS) stage, and overall survival in neuroblastoma. Higher taurine ssGSEA scores were generally observed in clinically more favorable subgroups and were associated with better overall survival in the discovery cohort. At the gene-set level, taurine score showed a strong positive correlation with hypoxia, but negative correlations with E2F targets, G2M checkpoint, MYC targets, and DNA repair, indicating a hypoxia/stress-associated biological context. Gene-level screening identified representative prognostic taurine-related candidates, from which an 8-gene candidate prognostic model was derived. In the discovery cohort, this model showed strong survival stratification and time-dependent discrimination, with 1-, 3-, and 5-year area under the receiver operating characteristic curve (AUC) values of 0.917, 0.906, and 0.913, respectively, and improved concordance when combined with clinical variables. External evaluation showed the strongest support in GSE181559, where Kaplan-Meier stratification was significant and AUCs were 0.701, 0.769, and 0.761 at 1, 3, and 5 years, respectively, whereas additional cohorts suggested inter-cohort heterogeneity and limited fixed-cutoff transferability. Biologically, the 8-gene score was positively associated with proliferative and metabolically active programs, including E2F, G2M, MYC, DNA repair, and glycolysis, but negatively associated with hypoxia and stromal/inflammatory programs. Integrative prioritization and patient-aware analyses further highlighted ADO, AURKA, GCLC, and MMP9 as four core candidate genes linked to a tumor-intrinsic proliferative component. A taurine metabolism-related transcriptional program is clinically relevant in neuroblastoma and, at the gene-set level, is associated with a hypoxia/stress-oriented biological background. In the discovery cohort, higher taurine ssGSEA scores tended to occur in clinically more favorable subgroups and were associated with better overall survival. From this candidate space, we derived an 8-gene candidate model that preferentially captures a proliferative and metabolically active risk component and receives partial external support. Further integrative prioritization and patient-aware analyses identified ADO, AURKA, GCLC, and MMP9 as core candidate genes for future functional validation.