Lactylation-Related Gene Biomarkers for Acute Myeloid Leukemia Prognosis

Seven hub lactylation-related genes (LSP1, MPO, GZMB, SPINK2, HLA-DRB1, HLA-DRA, POU2F2) were identified through integrative analysis of AML data from GEO and TCGA databases. Machine learning models, including LASSO-logistic and SVM-RFE, were used to validate these prognostic biomarkers. The study links these genes to immune pathways and potential therapeutic drug interactions.

Use Cases

• Validate the prognostic model using the GLM algorithm on the seven identified hub LRGs (LSP1, MPO, GZMB, SPINK2, HLA-DRB1, HLA-DRA, POU2F2).
• Analyze the relationship between immune cell infiltration, assessed via CIBERSORT, and the expression levels of key LRGs like GZMB and LSP1.
• Perform molecular docking simulations to screen drug candidates, such as (-)-Gallocatechin gallate, against target proteins like GZMB.
• Conduct single-gene GSEA to evaluate the functional enrichment of the seven hub LRGs in immune and inflammatory pathways.
• Assess pan-lactylation levels in AML cell lines, like Kasumi-1, under exogenous lactate or sodium oxamate treatment conditions.

Strengths

• Identifies seven specific hub lactylation-related genes (LSP1, MPO, GZMB, SPINK2, HLA-DRB1, HLA-DRA, POU2F2) as prognostic biomarkers.
• Analysis integrates data from two major public repositories, GEO and TCGA.
• Employs multiple machine learning algorithms (LASSO-logistic, SVM-RFE, Boruta) for gene selection and model validation.
• Includes experimental validation via qRT–PCR and immunohistochemistry on patient samples for GZMB and LSP1 expression.

Limitations

• The dataset is small at 155.1 KB, indicating limited raw data scope, likely containing summary results rather than primary sequencing data.
• Specific row and column counts for the underlying data tables are not provided, limiting reproducibility of exact analyses.
• Findings are specific to Acute Myeloid Leukemia and the lactylation mechanism, limiting generalizability to other cancers or biological processes.

Provenance

Source

Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases.

Collection Method

Integrative bioinformatics analysis using Seurat, limma, WGCNA, machine learning algorithms, and experimental validation (Western blot, qRT–PCR, IHC).

Freshness

Last updated March 23, 2026.