Fig 1.
Identification of key metabolic pathways in BCBM.
(A) OS differences between high and low activity groups of metabolic pathways in the Kaplan-Meier curves. Red indicates high-activity groups, blue indicates low-activity groups. (B) Correlation between 3 key metabolic pathways and immune infiltration scores. (C) Correlation between 3 key metabolic pathways and abundances of 28 immune cell types. Correlation coefficients were calculated using Spearman’s analysis. Red denotes negative correlation; blue denotes positive correlation. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig 2.
Metabolic pathway-related subtypes.
(A) Consensus heatmap of metabolic pathway clustering. (B-C) Heatmap and differential expression patterns of key metabolic pathways between C1 and C2. (D) Kaplan–Meier curves depict the OS differences between C1 and C2. (E) Differences of ImmuneScore, StromalScore, and TumorPurity between C1 and C2. (F) Differential abundances of 22 immune cell between C1 and C2. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig 3.
(A) Volcano plot of differentially expressed genes between C1 and C2. (B) Prognostic gene selection via elastic net modeling. (C) Consensus heatmap of MRGs. (D) Metabolic pathway activity heatmap across MRG1-3. (E) OS stratification by MRGs. (F) Differential ImmuneScore, StromalScore, and TumorPurity among MRGs. G: Varied abundances of 22 immune cell types across MRGs. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig 4.
Machine learning methods identify central prognostic genes.
(A-B) Reciprocal cumulative distribution curves determine model accuracy. (C) Top 10 genes among the six models. (D) Venn diagram identifies genes common to all six models. (E) ROC curve.
Fig 5.
Construction of a metabolism-related risk model.
(A) Kaplan-Meier curves of OS differences between high- and low-risk groups. (B) Differences in ImmuneScore, StromalScore, and TumorPurity between high- and low-risk groups. (C) Differences in the abundance of 28 immune cell types between high- and low-risk groups. (D) AUC values at 1, 3, and 5 years in the timeROC curve. (F) Drug sensitivity analysis between high- and low-risk groups. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig 6.
(A) UMAP and t-SNE visualizations depicting sample composition and distribution. (B) Expression of signature genes across 12 cell clusters. Dot size corresponds to the proportion of cells expressing specific markers, while color indicates the average expression level of each marker. (C) UMAP plot annotated with major cell types. (D) Differential ALDH1A1 expression across annotated cell types. (E) UMAP plot of cofactor/vitamin differentially expressed in each cell type.
Fig 7.
Cell-cell communication analysis.
(A) Cellular interaction networks and communication strength between high- vs. low-ALDH1A1 expressing cells. (B-C) Top two signaling pathway networks in high/low ALDH1A1 comparisons. (D) Interaction networks and communication strength between high- vs. low-Cofactor/Vitamin activity cells. (E-F) Top two signaling pathway networks in high/low Cofactor/Vitamin activity comparisons.
Fig 8.
(A) Macrophage Clustering. (B) M1/M2 Macrophage Markers. (C) ALDH1A1 Distribution in Macrophages. (D) Top Five Highly Variable Genes in the ALDH1A1-High Subpopulation. (E-F) Macrophage Trajectory and Pseudo-time Plots. (G) The relationship between the expression level of ALDH1A1 and pseudo-temporal sequence in the high ALDH1A1 subpopulation. (H) The distribution differences of the high-expression subpopulation of ALDH1A1 in the pseudo-temporal sequence.