【摘要】 目的 基于SEER数据建立四肢软骨肉瘤生存率列线图模型和外部数据验证及其临床意义。方法 提取2004至2016年美国国立癌症研究所SEER数据库所登记的四肢软骨肉瘤的临床数据，按7∶3比例随机分为建模组655例及内部验证组280例，采用单因素COX风险回归、Lasso回归、多因素COX风险回归分析确定影响四肢软骨肉瘤总生存率的独立危险因素，根据其生存率的独立危险因素构建列线图模型，利用一致性指数(C-index)、绘制校准曲线及时间依赖性ROC曲线检验模型的预测能力。收集2009至2020年在我院诊断为四肢软骨肉瘤患者27例作为外部验证组，再次利用时间依赖性ROC曲线检验模型的预测能力。结果 单因素分析结果显示：年龄、种族、性别、肿瘤分级、肿瘤分期、是否手术、淋巴结切除数、放疗、化疗、肿瘤直径是影响软骨肉瘤患者的预后因素。多因素COX回归分析结果显示：年龄、肿瘤分期、手术、淋巴结切除数、肿瘤直径是四肢软骨肉瘤独立危险因素。建模组C指数为0.848,其3年、5年、10年ROC曲线下面积分别为0.904、0.856、0.836。内部验证组C指数为0.835,其3年、5年、10年ROC曲线下面积分别为0.872、0.897、0.861。建立的列线图模型通过C指数、校准曲线、ROC曲线验证其有良好的预测效果。外部验证组3年、5年ROC曲线下面积分别为0.832、0.761。故所建立的列线图模型对中国本土病例也具有良好的预测价值。结论 建立四肢软骨肉瘤生存率列线图模型可以用于预测3、5年OS,可以使用于国内临床工作。更多还原

【Abstract】 Objective Columnar graph modeling of survival rates of chondrosarcoma of extremities based on SEER data and validation of external data and its clinical significance. Methods During 2004～2016,clinical data of chondrosarcoma of extremities registered in national cancer institute SEER database were extracted. In the ratio of 7∶3,they were randomly divided into modeling group(n=655) and internal validation group(n=280). Independent risk factors affecting overall survival of chondrosarcoma of extremities were determined using one-way COX risk regression, lasso regression, and multi-factor COX risk regression analysis. Columnar line graph models were constructed based on their independent risk factors for survival, and predictive ability of models was tested using consistency index(C-index),plotting calibration curves and time-dependent ROC curves. From 2009 to 2020,27 patients diagnosed with chondrosarcoma of extremities at our hospital were collected as an external validation group, and predictive ability of model using time-dependent ROC curves were again tested. Results Univariate analysis showed that age, race, gender, tumor grade, tumor stage, whether to operate, number of lymph nodes cleared, radiotherapy, chemotherapy, and tumor diameter were prognostic factors affecting patients with chondrosarcoma. Results of multi-factor COX regression analysis showed that age, tumor stage, surgery, number of lymph nodes removed, and tumor diameter were independent risk factors for chondrosarcoma of extremity bone. C-index of modeled group was 0.848,and its area under ROC curve at 3,5,and 10 years was 0.904,0.856,and 0.836,respectively. C-index of internal validation group was 0.835,and its area under ROC curve at 3,5,and 10 years was 0.872,0.897,and 0.861,respectively. Established column line graph model was validated by C-index, calibration curve, and ROC curves to verify that it had good predictive effect. In external validation group, the area under 3-year and 5-year ROC curves were 0.832 and 0.761,respectively. Therefore, established column line graph model also had good predictive value for local cases in China. Conclusion The established columnar line graph model for survival in chondrosarcoma of extremities cold be used to predict 3 and 5 year OS in domestic clinical work.更多还原