BACKGROUND

The prognosis of low-grade glioma (LGG) exhibits significant heterogeneity, and the optimal management strategy remains controversial. Therefore, identifying biomarkers associated with glioma prognosis is necessary.

METHODS

Hub genes were identified and a prognostic risk signature was constructed in the TCGA cohort using LASSO Cox regression. The predictive significance of this model was assessed and confirmed in the independent CGGA cohort. Single-cell transcriptional profiling characterized the hub gene expression patterns across cell types.

RESULTS

We screened four hub genes (SERPINE1, KIF2C, SLC16A1, FABP5). Patients were stratified into high- and low-risk groups using this model. The high-risk group had significantly lower survival rates in both TCGA and CGGA cohorts. The model demonstrated strong predictive power (1/2/3-year AUCs: 0.85/0.85/0.83 in TCGA; 0.73/0.71 for 2/3 years in CGGA). Functionally, the high-risk group exhibited activation of pro-tumorigenic pathways, showed higher levels of immune infiltration across multiple lymphocyte and myeloid subsets, and demonstrated higher sensitivity to drugs such as AZD5582 and AZD8055, evidenced by significantly lower IC50 values. At the single-cell level, FABP5 was highly expressed in myeloid cells, and FABP5 + tumor-associated macrophages (TAMs) were enriched in lipid metabolism and antigen presentation pathways. Pseudotime analysis suggested increasing FABP5 expression during TAMs differentiation.

CONCLUSIONS

The constructed 4-gene hypoxia-lactylation prognostic model can effectively predict survival risk in LGG patients. The high-risk group is characterized by activation of pro-tumorigenic pathways, high immune infiltration, and sensitivity to specific targeted drugs. FABP5 + TAMs participate in LGG progression by regulating lipid metabolism and inflammatory responses, representing a potential therapeutic target.