2026 年第 7 期 第 21 卷

经机器人辅助腹腔镜下肾部分切除术治疗的T1/T2期肾癌患者预后预测模型构建与验证

Construction and validation of a prognostic prediction model for T1/T2 stage renal cell carcinoma patients treated with robot-assisted laparoscopic partial nephrectomy

作者：车世庆黄媛媛白雅馨张薛张成伟

英文作者：Che Shiqing Huang Yuanyuan Bai Yaxin Zhang Xue Zhang Chengwei

单位：南京大学医学院附属鼓楼医院南京鼓楼医院泌尿外科，南京210008

英文单位：Department of Urology Nanjing Drum Tower Hospital the Affiliated Hospital of Nanjing University Medical School Nanjing 210008 China

关键词：肾癌；机器人辅助腹腔镜下肾部分切除术；预测模型

英文关键词：Renalcarcinoma;Robot-assistedlaparoscopicpartialnephrectomy;Predictionmodel

• 摘要：

• 目的 构建经机器人辅助腹腔镜下肾部分切除术（RALPN）治疗的T1/T2期肾癌患者预后的预测模型，并进行验证。方法 回顾性选取2018年8月至2019年7月在南京大学医学院附属鼓楼医院接受RALPN治疗的T1/T2期肾癌患者400例，构建预后的列线图预测模型，采用Bootstrap法进行内部验证，绘制校准曲线进行一致性验证。结果 400例患者随访5年脱落22例，有效随访378例，5年累积生存率为90.5%（342/378）、即生存组342例，36例（9.5%）死亡（死亡组）。Cox回归分析结果显示分化程度增高是RALPN患者预后的独立保护因素，年龄增高、有临床症状、世界卫生组织(WHO)/国际泌尿病理协会(ISUP)分级增高、T分期增高、有区域淋巴结转移是其独立危险因素（均P<0.05）。根据以上结果构建患者预后的风险预测模型，经Hosmer-Lemeshow拟合优度检验显示，χ2=6.762，自由度=8，P=0.578。列线图预测模型显示各因素对预后的影响大小依次是年龄、WHO/ISUP分级、分化程度、T分期、临床症状、区域淋巴结转移，当总分达到156分时，患者5年病死率可达85.0%。列线图模型的一致性指数为0.956，预测模型预测RALPN患者预后风险与实际发生风险一致性较好，Brier得分为0.076，校准斜率为0.820。结论 影响RALPN治疗的T1/T2期肾癌患者预后的因素包括分化程度、年龄、临床症状、WHO/ISUP分级、T分期、区域淋巴结转移，基于以上影响因素构建的预测模型具有良好的区分度、校准度。

• Objective To construct and validate a prognostic prediction model for T1/T2 stage renal carcinoma patients treated with robot-assisted laparoscopic partial nephrectomy (RALPN). Methods A total of 400 patients with T1/T2 stage renal carcinoma who underwent RALPN at the Affiliated Hospital of Nanjing University Medical School from August 2018 to July 2019 were retrospectively enrolled. A nomogram prediction model for prognosis was constructed, and internal validation was performed using the Bootstrap method, with calibration curves drawn for consistency verification. Results Among the 400 patients, 22 were lost to follow-up at 5 years, and 378 completed effective follow-up. The 5-year cumulative survival rate was 90.5%(342/378), including 342 in the survival group and 36 (9.5%) deaths in the death group. Cox regression analysis showed that increased differentiation degree was an independent protective factor for the prognosis of RALPN patients, while older age, presence of clinical symptoms, elevated World Health Organization (WHO)/International Society of Urological Pathology (ISUP) grade, advanced T stage, and regional lymph node metastasis were independent risk factors (all P<0.05). A prognostic risk prediction model was constructed based on the above results. Hosmer-Lemeshow goodness-of-fit test showed χ2=6.762, degree of freedom=8, P=0.578. The nomogram prediction model revealed that the influencing factors on prognosis ranked as age, WHO/ISUP grade, differentiation degree, T stage, clinical symptoms, and regional lymph node metastasis. When the total score reached 156, the 5-year mortality rate could reach 85.0%. The concordance index of the nomogram model was 0.956, indicating good consistency between predicted prognostic risk and actual risk in RALPN patients. The Brier score was 0.076, and the calibration slope was 0.820. Conclusion Prognostic factors for T1/T2 stage renal carcinoma patients treated with RALPN include differentiation degree, age, clinical symptoms, WHO/ISUP grade, T stage, and regional lymph node metastasis. The prediction model constructed based on these factors shows favorable discrimination and calibration.