Development and validation of a disulfidptosis-related genes signature for predicting outcomes and immunotherapy in acute myeloid leukemia.

BACKGROUND

Acute myeloid leukemia (AML) is a hematopoietic malignancy with poor outcomes and high recurrence. Disulfidptosis, a novel form of programmed cell death driven by aberrant disulfide bonds and F-actin collapse, provides insights into cancer progression and treatment.

METHODS

We investigated the correlation network and prognostic values of disulfidptosis-related genes (DRGs) in AML. Unsupervised clustering was performed to reveal distinct disulfidptosis-related AML subtypes. We implemented the differential analysis and enrichment analysis to explore the difference of the distinct subtypes in biological processes. Least absolute shrinkage and selection operator (LASSO) Cox model was used to generate a disulfidptosis-related signature. We employed the ESTIMATE, CIBERSORT, and scRNA analyses to assess the tumor microenvironment of AML. Moreover, experiments validated the functions of PTPN6 and CSK in OCI-AML2 cells.

RESULTS

We identified 10 prognostic DRGs and revealed two disulfidptosis subtypes. DRGs significantly affected immune processes like interferon-gamma response and MHC class II antigen presentation. LASSO algorithm was implemented to established a 6-gene signature (HLA-DRB5, CCDC124, PTPN6, HLA-DMA, CSK, ISG15) that predicted prognosis in two validation cohorts more robustly than other signatures. Disulfidptosis was correlated with tumor microenvironment immune cells, especially monocytes. The two risk subgroups differed significantly in susceptibilities of multiple chemotherapy drugs, indicating disulfidptosis as a potential therapeutic target. Knockdown of PTPN6 and CSK inhibited the proliferation of AML cells and increased apoptosis.

CONCLUSIONS

Our study provides insights into DRG prognoses and immunomodulation, establishing a robust 6-gene risk model for predicting AML outcomes that may enhance precision medicine and treatment strategies.