Identification of disulfidptosis-related long non-coding RNA signature to predict the prognosis, immunotherapy, and chemotherapy options in acute myeloid leukemia.

Abstract

Background: Disulfidptosis, a recently identified programmed cell death mechanism, has emerged as a critical regulator in tumorigenesis and demonstrates significant prognostic value across multiple cancer types. However, the prognostic significance of disulfidptosis-related long non-coding RNAs (DRLs) in acute myeloid leukemia (AML) and their functional implications in the tumor immune microenvironment (TIME) remain poorly characterized. Furthermore, the expression patterns and regulatory mechanisms of DRLs in AML require systematic investigation to elucidate their potential clinical applications. The study aims to investigate the prognostic and immunotherapeutic implications of DRLs in AML.

Methods: RNA sequencing and clinical data for AML samples, as well as genotype-tissue expression (GTEx) normal bone marrow samples, were sourced from the University of California Santa Cruz (UCSC) database. Initially, DRLs were identified using Pearson correlation analysis. Subsequently, univariate Cox proportional hazards regression analysis was employed to identify long non-coding RNAs (lncRNAs) associated with prognosis. Key prognostic biomarkers were then selected through least absolute shrinkage and selection operator (LASSO) regression, stepwise Cox regression (StepCox), CoxBoost, and random survival forest (RSF) methods. A prognostic model was developed utilizing multivariate Cox regression analysis, and correlations between DRL risk scores, the AML immune microenvironment, and therapeutic agents were predicted. Furthermore, the expression levels of these DRLs in AML cell lines were validated by quantitative reverse transcription-polymerase chain reaction (RT-PCR).

Results: We identified eight pivotal DRLs and developed a DRLs-based risk model (DRLs-RM). Patients classified in the low-risk cohort exhibited prolonged survival compared to those in the high-risk cohort. Multivariate Cox proportional hazards analysis demonstrated that DRL risk scores function as an independent prognostic biomarker for AML. Enrichment analysis revealed that DRL risk scores correlate with apoptotic pathways and NADPH oxidoreductase activity. Furthermore, DRL risk scores showed significant associations with the AML immune microenvironment, including elevated expression of various immune checkpoint molecules and human leukocyte antigen (HLA) genes in the high-risk group. Drug sensitivity profiling indicated that high-risk patients exhibit increased sensitivity to agents such as axitinib and cyclin-dependent kinase 9 (CDK9) inhibitors.

Conclusions: The prognostic model incorporating eight DRLs demonstrates high accuracy and reliability in predicting survival outcomes for AML patients, thereby identifying potential therapeutic targets for future AML treatment strategies.