肿瘤微环境（TME）在急性髓系白血病（AML）的发生、发展中发挥着重要作用。单细胞测序使研究人员能够探索TME亚群与肿瘤预后之间的相关性，区分肿瘤细胞耐药亚群的存在，并揭示AML细胞异质性的复杂性。我们使用公共数据库中的骨髓免疫细胞富集分析来筛选预后基因，构建预后模型，并在独立的外部数据集和患者样本上验证其预后意义。共获得18,251个单细胞，建立CCL5、ETLS2、IL2RA等10个关键基因的预后评分模型。将AML病例分为高危组和低危组。低风险组的生存率高于高风险组。 TCGA和GEO训练集中1年、3年和5年生存曲线的曲线下面积（AUC）分别大于0.8和0.6，表明预测有效。该模型的预后功效在多个验证集上得到了证实。结果表明，与对照相比，AML 样本中 ETS2、CCL5 和 IL2RA 的表达增加，这与总生存期 (OS) 降低相关。这种基于肿瘤免疫浸润的预后评分模型为开发复发/难治性 AML 的新治疗策略提供了参考。© 2024。作者。

The tumor microenvironment (TME) plays an important role in the occurrence and progression of Acute Myeloid Leukemia (AML). Single-cell sequencing has enabled researchers to explore the correlation between TME subgroups and tumor prognosis, distinguish the existence of drug-resistant subgroups of tumor cells, and unravel the complexity of the AML cellular heterogeneity. We used bone marrow immune cell enrichment analysis from public databases to screen prognostic genes, construct prognostic models, and validate their prognostic significance on independent external datasets and patient samples. A total of 18,251 single cells were obtained to establish prognostic scoring models for 10 key genes including CCL5, ETLS2, and IL2RA.The AML cases were divided into two groups: high-risk and low-risk. The low-risk group exhibited a higher survival rate than the high-risk group. The areas under curves (AUC) of 1-, 3- and 5-year survival curves in the TCGA and GEO training sets were greater than 0.8 and 0.6, respectively, indicating effective prediction. The model's prognostic efficacy was confirmed across multiple validation sets. It demonstrated increased expression of ETS2, CCL5, and IL2RA in AML samples compared to controls, which was associated with decreased overall survival (OS). This prognostic scoring model based on tumor immune infiltration provides a reference for developing novel treatment strategies for recurrent/refractory AML.© 2024. The Author(s).