N6-甲基腺嘌呤（m6A）已被证实在急性髓性白血病（AML）的进展中起到关键作用。至今，m6A相关基因在AML中的分型和预后预测意义尚不清楚。我们从The Cancer Genome Atlas （TCGA-LAML）、可应用治疗研究产生有效治疗（TARGET-AML）和基因表达谱（GEO，GSE71014）数据库中收集了AML患者的测序数据。通过limma软件包去除了TCGA-LAML和TARGET-AML的批次效果，并得到了AML队列的聚合样本。通过一致性聚类方法，使用基于23个m6A相关调控因子的共识聚类方法，识别了AML队列中与m6A相关的模型。通过差异表达分析和单变量Cox分析确定影响AML总体生存（OS）的m6A相关差异表达基因（m6ARDEGs），并使用基于共识的聚类方法评估AML分子亚型。采用LASSO和多变量Cox分析来获取优化的m6ARDEGs来构建m6A预后风险评分（m6APR_Score）。我们根据Kaplan Meier（K-M）和受试者工作特征曲线（ROC）评估模型的稳健性。此外，还研究了不同m6A修饰模式、分子亚型和m6APR_Score分组中免疫细胞浸润的丰度。最后，构建了一个预测AML患者OS的策略评分图。使用定量实时聚合酶链反应（RT-qPCR）和细胞计数试剂盒（CCK-8）试验验证了AML细胞中m6APR_Score中的基因。在AML队列中确定了m6A模型（m6AM1、m6AM2、m6AM3）和分子亚型（C1、C2、C3），展示了不同的预后和免疫反应性。我们同时发现了AML的新的预后生物标志物，如CD83、NRIP1、ACSL1、METTL7B、OGT和C4orf48。将AML患者分为高m6APR_Score组和低m6APR_Score组，后者的预后比前者好。无论是AML队列还是验证队列GSE71014都表现出了出色的预测能力。最后，策略评分图准确地预测了患有AML的患者的生存情况。此外，决策曲线显示策略评分图和m6APR_Score都表现出出色的预测性。通过体外实验证实，在AML细胞中NRIP1、ACSL1、METTL7B和OGT的mRNA表达升高，而CD83和C4orf48的mRNA表达降低。在抑制CD83后，U937和THP-1细胞系的存活率显著增加，而siMETTL7B则产生了相反的结果。我们的研究证明，m6APR_Score以及CD83、NRIP1、ACSL1、METTL7B、OGT和C4orf48可能为AML患者提供新的有前景的预后支持。© 2023. BioMed Central Ltd., 由Springer Nature旗下的一部分。

N6-methyladenosine (m6A) has been confirmed to function critically in acute myeloid leukemia (AML) progression. Hitherto, the subtyping and prognostic predictive significance of m6A-correlated genes in AML is unclear.From The Cancer Genome Atlas (TCGA-LAML), Therapeutically Applicable Research to Generate Effective Treatments (TARGET-AML) and Gene Expression Omnibus (GEO, GSE71014) databases, we collected the sequencing data of AML patients. The batch effect was removed via limma package for TCGA-LAML and TARGET-AML, and the aggregated samples were AML cohorts. Samples in the AML cohort identified m6A models in AML by consensus clustering based on 23-m6A-related modulators. M6A-related differentially expressed genes (m6ARDEGs) influencing the overall survival (OS) of AML were determined by performing differential expression analysis and univariate COX analysis, and consensus-based clustering was utilized to access AML molecular subtypes. LASSO and multivariate COX analyses were performed to obtain the optimized m6ARDEGs to construct the m6A Prognostic Risk Score (m6APR_Score). Whether the model was robust was evaluated according to Kaplan-Meier (K-M) and receiver operator characteristic (ROC) curves. Further, the abundance of immune cell infiltration was explored in different m6A modification patterns and molecular subtypes and m6APR_Score groupings. Finally, nomogram was constructed to predict OS in AML. Quantitative real-time polymerase chain reaction (RT-qPCR) and cell counting kit-8 (CCK-8) assay were used to validate the genes in m6APR_Score in AML cells.The m6A models (m6AM1, m6AM2, m6AM3) and molecular subtypes (C1, C2, C3) were identified in the AML cohort, exhibiting different prognosis and immunoreactivity. We recognized novel prognostic biomarkers of AML such as CD83, NRIP1, ACSL1, METTL7B, OGT, and C4orf48. AML patients were grouped into high-m6APR_Score and low-m6APR_Score groups, with the later group showing a better prognosis than former one. Both the AML cohort and the validation cohort GSE71014 demonstrated excellent prediction. Finally, the nomogram accurately predicted the survival of patients suffering from AML. Further, the decision curves showed that both nomogram and m6APR_Score showed excellent prediction. It was confirmed in vitro experiments that mRNA expressions of NRIP1, ACSL1, METTL7B and OGT were elevated, while CD83 and C4orf48 mRNA expressions downregulated in AML cells. A significant increase in the viability of U937 and THP-1 cell lines after inhibition of CD83, while siMETTL7B had contrast results.Our study demonstrated that m6APR_Score and CD83, NRIP1, ACSL1, METTL7B, OGT, and C4orf48 potentially provided novel and promising prognostic support for AML patients.© 2023. BioMed Central Ltd., part of Springer Nature.