A prognostic biomarker of disulfidptosis constructed by machine learning framework model as potential reporters of pancreatic adenocarcinoma

Abstract

Background: Pancreatic adenocarcinoma (PAAD), known for its high lethality, has not been thoroughly explored in terms of its mechanisms and patterns of immune infiltration. Disulfidptosis, a newly identified mode of cell death, is likely associated with tumorigenesis and progression but remains poorly understood in PAAD at the genetic and mechanistic levels.

Methods: Sixteen PAAD samples from the GSE154778 scRNA-seq dataset were subjected to single-cell analysis. Disulfidptosis grouping and scores were established across various immune cell populations. Using the TCGA-PAAD database, LASSO regression was employed to construct prognostic markers linked to disulfidptosis. The performance of this model was assessed in both training and independent validation cohorts. Subsequent analyses explored the correlations between disulfidptosis scores, immune infiltration, and drug sensitivity. Cellular experiments further confirmed the significant positive relationship of the gene MET with disulfidptosis and its role in influencing the invasion and metastasis of PAAD.

Results: WGCNA identified Disulf-High and Disulf-Low as modules strongly correlated with disulfidptosis. Five prognostically significant genes were selected to construct prognostic models. Survival analysis demonstrated that the disulfidptosis-related biological model successfully achieved prognostic stratification in PAAD patients. Additionally, the disulfidptosis score was significantly correlated with both immune infiltration and drug sensitivity. Knockdown of the MET gene substantially inhibited cell multiplication and cell cycle progression in two PAAD cell lines, effects potentially induced by the activation of the PI3K/AKT signalling pathway in the tumour.

Conclusion: Key genes associated with disulfidptosis significantly correlate with immune infiltration and the development of PAAD. Biomarkers based on disulfidptosis present potential avenues for novel therapies and clinical treatments in PAAD.

Keywords: Disulfidptosis; Machine learning; Pancreatic adenocarcinoma; Progosis biomarker.