Apolipoprotein C1 and C-X-C motif chemokine receptor 4 as myeloid-associated candidate genes in glioblastoma: An integrative transcriptomic analysis

Background Tumor-associated myeloid cells, including resident microglia and recruited macrophages, are major components of the tumor microenvironment in glioblastoma (GBM). This study aimed to identify myeloid-associated candidate genes linked to GBM through an integrated single-cell and bulk transcriptomic analysis. Methods Single-cell RNA sequencing data from six GBM samples in GSE221534 were used for provisional cell-state annotation, CopyKAT-based copy-number inference, CellChat analysis, and high-dimensional weighted gene co-expression network analysis (hdWGCNA). Myeloid-associated module genes were intersected with differentially expressed genes identified from the pan-glioma GSE4290 microarray cohort, which comprised 157 glioma samples and 23 epilepsy-derived non-tumor brain samples. Least absolute shrinkage and selection operator (LASSO) regression and support vector machine-recursive feature elimination (SVM-RFE) were used for feature selection. CIBERSORT, ESTIMATE, and exploratory DSigDB drug-signature analyses were subsequently performed. Results Eleven putative cell populations or states and five hdWGCNA modules were identified. CopyKAT showed mixed copy-number states within the provisionally annotated glial-like populations. Modules M2 and M3 showed relatively greater activity in the annotated microglial/myeloid cluster. In the pan-glioma GSE4290 comparison, 2427 differentially expressed genes, including 953 upregulated and 1474 downregulated genes, were identified, and 17 overlapped with M2/M3 genes. Ten-fold cross-validation of LASSO and SVM-RFE yielded eight shared candidate genes: MALAT1, APOE, CEBPD, APOC1, NR4A2, CXCR4, C1QA, and FCGRT. Across the full GSE4290 cohort, APOC1 and CXCR4 were positively correlated with CIBERSORT-inferred M2 macrophage fractions (Spearman ρ = 0.621 and 0.600, respectively; adjusted P = 7.60 × 10−19 and 2.36 × 10−17, respectively, with adjusted P < 0.05 used as the study-wide significance threshold). Pioglitazone was among nine significant DSigDB terms (adjusted P = 0.00988) and overlapped with APOC1 and CXCR4. Discussion The analysis nominated APOC1 and CXCR4 as myeloid-associated candidate genes rather than cell-type-specific or clinically validated biomarkers. The pan-glioma bulk, immune-deconvolution, CellChat, and drug-signature findings are exploratory and require validation in independent GBM cohorts. Conclusion APOC1 and CXCR4 are candidate genes associated with GBM-related myeloid programmes. Their cellular localisation, prognostic relevance, and therapeutic implications require independent and experimental validation.