上尿路上皮癌(UTUC)是一种异质性较高的恶性肿瘤,占尿路上皮肿瘤的5%~10%,其预后受到患者特征、肿瘤病理特性及治疗方式等多种因素的综合影响。尽管根治性肾输尿管切除术(RNU)是治疗UTUC的金标准,但术后复发率和远期生存率差异显著。构...上尿路上皮癌(UTUC)是一种异质性较高的恶性肿瘤,占尿路上皮肿瘤的5%~10%,其预后受到患者特征、肿瘤病理特性及治疗方式等多种因素的综合影响。尽管根治性肾输尿管切除术(RNU)是治疗UTUC的金标准,但术后复发率和远期生存率差异显著。构建个体化的预后模型对于优化临床决策具有重要意义。近年来,诺模图、机器学习驱动模型、分子生物标志物模型及联合影像学与临床数据的多变量模型在UTUC预后预测中逐渐应用。其中,诺模图凭借直观性和高整合性成为临床预测的常用工具,机器学习模型在处理多模态数据方面表现出优势,分子生物标志物模型揭示了疾病的分子机制,而联合影像学模型通过融合影像和临床数据进一步提升了预测精准性。然而,现有模型的普适性和动态预测能力仍面临挑战,模型依赖于高质量的大规模数据,而临床实践中数据获取和整合存在难点。未来研究应聚焦于多中心、大样本的前瞻性研究以验证模型的可靠性,同时深入探索UTUC的分子机制,开发新的分子标志物,优化辅助治疗的适应症,并推动影像学技术与分子诊断手段的结合,为UTUC患者的精准医学和个体化治疗提供更可靠的工具和方法。Upper tract urothelial carcinoma (UTUC) is a highly heterogeneous malignancy, accounting for 5%~10% of urothelial tumors. Its prognosis is influenced by a combination of patient characteristics, tumor pathology, and treatment strategies. Despite radical nephroureterectomy (RNU) being the gold standard treatment for UTUC, significant variability in postoperative recurrence rates and long-term survival outcomes exists. Developing individualized prognostic models is crucial for optimizing clinical decision-making. Recently, nomograms, machine learning-based models, biomarker-driven molecular models, and multivariate models integrating imaging and clinical data have been increasingly utilized in UTUC prognostic prediction. Among these, nomograms have become widely used for their intuitive and integrative capabilities, machine learning models excel in handling multimodal data, biomarker-driven models uncover the molecular mechanisms of disease, and imaging-based models improve prediction accuracy by combining radiological and clinical data. However, existing models face challenges regarding generalizability and dynamic prediction capabilities, as they often rely on large-scale, high-quality datasets, which are difficult to obtain and integrate in clinical practice. Future research should focus on conducting multicenter, large-scale prospective studies to validate model reliability, exploring molecular mechanisms of UTUC, developing novel biomarkers, optimizing indications for adjuvant therapies, and promoting the integration of advanced imaging.展开更多
文摘上尿路上皮癌(UTUC)是一种异质性较高的恶性肿瘤,占尿路上皮肿瘤的5%~10%,其预后受到患者特征、肿瘤病理特性及治疗方式等多种因素的综合影响。尽管根治性肾输尿管切除术(RNU)是治疗UTUC的金标准,但术后复发率和远期生存率差异显著。构建个体化的预后模型对于优化临床决策具有重要意义。近年来,诺模图、机器学习驱动模型、分子生物标志物模型及联合影像学与临床数据的多变量模型在UTUC预后预测中逐渐应用。其中,诺模图凭借直观性和高整合性成为临床预测的常用工具,机器学习模型在处理多模态数据方面表现出优势,分子生物标志物模型揭示了疾病的分子机制,而联合影像学模型通过融合影像和临床数据进一步提升了预测精准性。然而,现有模型的普适性和动态预测能力仍面临挑战,模型依赖于高质量的大规模数据,而临床实践中数据获取和整合存在难点。未来研究应聚焦于多中心、大样本的前瞻性研究以验证模型的可靠性,同时深入探索UTUC的分子机制,开发新的分子标志物,优化辅助治疗的适应症,并推动影像学技术与分子诊断手段的结合,为UTUC患者的精准医学和个体化治疗提供更可靠的工具和方法。Upper tract urothelial carcinoma (UTUC) is a highly heterogeneous malignancy, accounting for 5%~10% of urothelial tumors. Its prognosis is influenced by a combination of patient characteristics, tumor pathology, and treatment strategies. Despite radical nephroureterectomy (RNU) being the gold standard treatment for UTUC, significant variability in postoperative recurrence rates and long-term survival outcomes exists. Developing individualized prognostic models is crucial for optimizing clinical decision-making. Recently, nomograms, machine learning-based models, biomarker-driven molecular models, and multivariate models integrating imaging and clinical data have been increasingly utilized in UTUC prognostic prediction. Among these, nomograms have become widely used for their intuitive and integrative capabilities, machine learning models excel in handling multimodal data, biomarker-driven models uncover the molecular mechanisms of disease, and imaging-based models improve prediction accuracy by combining radiological and clinical data. However, existing models face challenges regarding generalizability and dynamic prediction capabilities, as they often rely on large-scale, high-quality datasets, which are difficult to obtain and integrate in clinical practice. Future research should focus on conducting multicenter, large-scale prospective studies to validate model reliability, exploring molecular mechanisms of UTUC, developing novel biomarkers, optimizing indications for adjuvant therapies, and promoting the integration of advanced imaging.
