【摘要】 目的 通过综合多组学分析与人工神经网络(ANN)划分肾上腺皮质癌(ACC)的精准分子亚型并建立预后预测模型，探索不同亚型与免疫治疗反应相关的生物学特征，为ACC的准确预后评估及个体化治疗提供参考。方法 整合2012年1月1日—2022年12月31日四川大学华西医院泌尿外科收治的44例中国ACC患者的多组学数据(包括基因组、转录组和临床特征),采用10种不同聚类算法进行共识聚类，识别ACC分子亚型。随后将聚类结果整合入ANN模型中，构建ANN驱动的预后指数(ANPI)模型，并将其用于患者分层与预后评估。结果 共识别出3个具有显著预后差异的ACC分子亚型(CS1～3),其中CS1亚型具有最差的预后表现。基于筛选出的20个核心基因构建ANPI模型。ANPI能够有效区分不同预后风险的患者群体：ANPI低风险组具有显著更佳的总生存率，并表现出“热肿瘤”免疫特征，提示更高的免疫治疗获益可能性；而ANPI高风险组则呈现“冷肿瘤”免疫特征，预后较差且对免疫治疗反应不佳。结论 通过综合多组学分析构建中国人群ACC的分子分型图谱，明确与免疫治疗反应相关的关键生物特征。ANPI模型在早期预后预测和免疫治疗获益人群识别方面具有良好性能，为ACC的精准医疗提供了理论基础与临床工具，具有重要的转化应用价值。更多还原

【Abstract】 Objective To explore the biological characteristics associated with different subtypes and the response to immunotherapy by integrating multiomics analysis and artificial neural networks(ANN) to delineate the precise molecular subtypes of adrenocortical carcinoma(ACC) and establish a prognostic prediction model, in order to provide reference for the accurate prognosis assessment and individualized treatment of ACC.Methods The multiomics data of 44 Chinese ACC patients admitted to the Department of Urology, West China Hospital of Sichuan University during Jan.1,2012 and Dec.31,2022 were integrated, including genomic, transcriptomic and clinical features.Ten different clustering algorithms were employed for consensus clustering to identify robust molecular subtypes.The results were then incorporated into an ANN model to construct an ANN-driven prognostic index(ANPI) for patient stratification and survival prediction.Results Three distinct molecular subtypes(cancer subtypes, CS1-3) with significantly different prognoses were identified, among which CS1 exhibited the poorest survival outcomes.A set of 20 core genes was selected to form the basis of the ANPI model.ANPI effectively stratified patients into high-and low-risk groups: patients in the low-ANPI group had significantly better overall survival and exhibited “hot tumor” immune phenotypes, suggesting greater benefits from immunotherapy.In contrast, high-ANPI patients had worse prognoses and displayed “cold tumor” characteristics with weaker immunotherapy responses.Conclusion Our integrative multiomics analysis illustrated the molecular landscape of ACC in the Chinese population and uncovered the key immune-related features linked to clinical outcomes.The ANPI model demonstrated strong performance in prognostic prediction and immunotherapy response assessment, offering a valuable tool for precision oncology and clinical decision-making.更多还原