Abstract: Objective Neutrophil extracellular traps （NETs） can be stimulated by various factors, including drug perfusion and tumor cell stimulation, thereby influencing the prognosis of cancer patients. However, the prognostic impact and key functional genes of NETs in the recurrence of non-muscle-invasive bladder cancer （NMIBC） remain unclear. This study aims to identify critical NETs-related genes associated with NMIBC recurrence and provide a reliable predictive tool for clinical recurrence assessment. Methods Transcriptomic data and clinical information from bladder cancer patients were obtained from the GEO database （GSE13507, GSE128959, GSE19423, GSE154261, GSE31684, GSE169455）, and somatic copy number variation （CNV） data were retrieved from TCGA. Using machine learning algorithms and weighted gene co-expression network analysis （WGCNA）, we identified 153 NETs-related genes and constructed a recurrence prediction score （NRG）, which was validated in a training cohort. We comprehensively analyzed the impact of this score on gene expression, the immune microenvironment, and functional pathways in bladder cancer. Additionally, we explored potential sensitivities to NRG-associated small-molecule compounds to identify therapeutic targets for clinical intervention. Results This study identified three NETs-related genes （G0S2, CCL5, and CLEC7A） as independent prognostic predictors for postoperative recurrence in NMIBC patients. The NRG score effectively predicted recurrence outcomes in the training cohort, demonstrating diagnostic AUC values of 0.671 and 0.645 in two independent NMIBC datasets, with significant prognostic stratification （P=0.039）. Genomic and immune infiltration analyses revealed that high-NRG patients exhibited more frequent PIK3CA mutations and increased infiltration of immunosuppressive cell subsets. Functional enrichment indicated hyperactivation of immune checkpoint pathways in high-NRG cases. Drug sensitivity analysis suggested that targeting NRG may reduce recurrence risk by inhibiting the PI3K-mTOR and ERK signaling axes, providing potential therapeutic strategies for NMIBC. Conclusion This study established a NETs-derived recurrence prediction signature （NRG） for NMIBC and elucidated its immunomodulatory effects within the tumor microenvironment, functional pathway alterations, and potential small-molecule therapeutic targets.

Key words: bladder cancer, machine learning, neutrophil extracellular traps, therapeutic targets

摘要： 目的 中性粒细胞胞外诱捕网（NETs）可由药物灌注和肿瘤细胞刺激等多种因素刺激中性粒细胞释放从而影响肿瘤患者预后。然而NETs在非肌层浸润性膀胱癌（NMIBC）复发过程中的预后影响和关键作用基因尚未明确,因此本研究旨在鉴定影响NMIBC复发的关键NETs相关基因评分为临床提供可靠的复发预测工具。方法 从GEO数据库（GSE13507、GSE128959、GSE19423、GSE154261、GSE31684、GSE169455）收集膀胱癌测序数据及临床信息,TCGA数据库获取体细胞拷贝数变异数据。通过机器学习算法和加权基因共表达网络（WGCNA）筛选153个NETs相关基因,构建复发预测评分（NRG）,并在训练集中验证其预测效能。综合分析了该评分对膀胱癌患者基因、免疫微环境及功能通路的影响,同时探索了NRG相关的小分子化合物潜在敏感性为临床治疗提供靶点。结果 本研究鉴定出3个NETs相关的NMIBC复发预测基因（G0S2、CCL5、CLEC7A）,对NMIBC患者术后复发具有独立的危险预测效能。NRG在训练集中可有效预测NMIBC患者复发结局,在对两个独立的NMIBC数据集复发的诊断效能分别可达到0.671和0.645并可有效区分复发预后（P=0.039）。基因组和免疫亚群浸润分析提示高表达NRG评分的患者伴随更多PIK3CA突变及免疫抑制细胞浸润。功能分析表明,NRG高表达病例中PD-1等多种免疫抑制调控通路高度富集。抗癌药物敏感性分析表明NRG的靶向可能通过阻断PI3K-mTOR和ERK信号轴等通路降低NMIBC患者术后复发风险。结论 本研究建立了NETs相关的NMIBC复发预测评分-NRG,并全面探讨了NRG表达在肿瘤微环境中的免疫调节影响、功能通路变化和可能有效的小分子靶向药物。

关键词: 膀胱癌, 机器学习, 中性粒细胞胞外诱捕网, 药物靶点