目的 基于免疫相关基因构建神经母细胞瘤预后模型，探讨其在神经母细胞瘤中的作用，并预测挖掘潜在的干预中药。方法 通过limma包对TARGET-NBL数据集中高危患者组和非高危组表达谱数据进行差异分析。差异基因与ImmPort数据库中的免疫相关基因取交集，对差异免疫相关基因进行富集分析。应用单因素/多因素Cox回归分析、Lasso回归分析进一步筛选预后相关免疫差异基因，据此建立预后模型。采用CIBERSORT和单样本基因集富集分析（single-sample gene-set enrichment analysis，ssGSEA）评估高风险评分组和低风险评分组样本中免疫细胞的丰度。通过CTD和COREMINE数据库预测与预后模型中基因高度相关的化合物和中药。结果 在高危患者中147个差异免疫相关基因显著富集在Janus激酶（Janus kinase，JAK）/信号转导和转录激活因子（signal transducer and activator of transcription，STAT）、Ras、B细胞受体、环磷酸腺苷（cyclic adenosine monophosphate，cAMP）、磷脂酰肌醇-3-羟激酶（phosphatidylinositol-3-hydroxykinase，PI3K）/蛋白激酶B（protein kinase B，Akt）、Rap1、丝裂原活化蛋白激酶（mitogen-activated protein kinase，MAPK）、转化生长因子-β（transforming growth factor-β，TGF-β）等多个信号通路。单因素/多因素Cox回归分析显示促甲状腺激素β亚基（thyroid stimulating hormone subunit beta，TSHB），肽脱甲酰基酶、线粒体（peptide deformylase, mitochondrial，PDF），血管生成素样蛋白3（angiopoietin like 3，ANGPTL3），前列腺素E受体1（prostaglandin E receptor 1，PTGER1）是高危神经母细胞瘤预后的独立影响因素。基于Lasso分析构建的免疫预后模型中，高风险评分组较低风险组评分组预后差（HR＝4.49，95% CI：2.40～8.38，P＜0.001)，Calibration校正曲线和决策曲线也证明了该模型具有较好的预后效能和一定的临床实用性。CIBERSORT和ssGSEA免疫分析均显示CD8⁺T细胞在预后模型低风险评分组中浸润水平较高风险评分组显著升高（P＜0.05），且与风险评分呈显著负相关（P＜0.05）。此外，发现双酚A（bisphenol A）可同时调控ANGPTL3、PDF、PTGER1、甲状旁腺激素受体2（parathyroid hormone receptor 2，PTH2R）、TSHB、神经生长因子诱导蛋白（nerve growth factor inducible，VGF）的表达，进而抑制神经母细胞瘤的进展。基于免疫预后模型基因靶点预测得到菟丝子、茶树根等130味核心干预中药，药味以苦、甘、辛为主，药性以寒、温为主，归经以肝、肺、脾、肾为主，功效以清热、补虚为主。结论 在高危型神经母细胞瘤中，免疫相关基因ANGPTL3、PDF、PTGER1、PTH2R、TSHB、VGF构建了具有较好预后效能和一定临床实用性的预后模型。通过整合多个数据库预测了作用于预后模型基因的药物，为高危型神经母细胞瘤的治疗提供新的方向。

Objective To construct a neuroblastoma prognostic model based on immune-related genes, explore its role in neuroblastoma, and predict potential intervention Chinese medicine. Methods Limma package was used to analyze the difference between the expression profile data of high-risk patients and non-high-risk patients in the TARGET-NBL dataset. Differential genes were overlapped with immune-related genes from ImmPort database, and the resulting differential immune-related genes were enriched and analyzed. Univariate/multifactor Cox regression analysis and Lasso regression analysis were used to further screen prognostic-related immune differential genes, and the prognostic model was established. CIBERSORT and single-sample gene-set enrichment analysis (ssGSEA) were used to assess the abundance of immune cells in high risk score and low risk score samples. CTD and COREMINE databases were used to predict the compounds and Chinese medicines that highly interact with genes in the prognostic model. Results In high-risk patients, 147 differentially immune-related genes were significantly enriched in Janus kinase/signal transducer and activator of transcription (JAK/STAT), Ras, B cell receptor, cyclic adenosine monophosphate (cAMP), phosphatidylinositol-3-hydroxykinase/protein kinase B (PI3K/Akt), Rap1, mitogen-activated protein kinase (MAPK), transforming growth factor-β (TGF-β) and other signaling pathways. Univariate/multivariate Cox regression analysis showed that thyroid stimulating hormone subunit beta (TSHB), peptide deformylase, mitochondrial (PDF), angiopoietin like 3 (ANGPTL3), prostaglandin E receptor 1 (PTGER1) were independent prognostic factors of high-risk neuroblastoma. In the immune prognosis model constructed based on Lasso analysis, the high risk group had worse prognosis than the low risk group (HR = 4.49, 95% CI: 2.40—8.38, P < 0.001). Calibration curve and decision curve also prove that the model had good prognostic efficacy and certain clinical practicability. Both CIBERSORT and ssGSEA results showed that CD8⁺T cells infiltration level was significantly increased in the low-risk group of prognostic model (P < 0.05) and significantly negatively correlated with risk score (P < 0.05). In addition, it was found that bisphenol A can simultaneously regulate the expression of ANGPTL3, PDF, PTGER1, parathyroid hormone receptor 2 (PTH2R), TSHB and nerve growth factor inducible nerve growth factor inducible (VGF), thus inhibiting the progression of neuroblastoma. Based on the gene target of immune prognosis model, 130 core traditional Chinese medicines such as Tusizi (Cuscutae Semen) and Chashugen (root of Camellia sinensis) were predicted, characterized by bitter, sweet and pungent tastes. The medicinal properties primarily focused on cold and warm, while the liver, lung, spleen and kidney meridians were the main channels affected, the function weremainly clearing heat and tonifying deficiency. Conclusion In high-risk neuroblastoma, immune-related genes ANGPTL3, PDF, PTGER1, PTH2R, TSHB and VGF construct a prognostic model with good prognostic efficacy and certain clinical practicability. By integrating multiple databases to predict drugs that act on prognostic model genes, it provides a new direction for the treatment of high-risk neuroblastoma.

R285

北京“十四五”中医药重点专科项目（领超类和示范类）（BJZKLC005）