基于SEER数据库的壶腹周围癌预后预测模型的构建

Construction of a prognostic prediction model for periampullary cancer based on SEER database

目的分析影响壶腹周围癌(periampullary carcinoma)预后的危险因素,建立壶腹周围癌预后预测模型。方法回顾性分析美国SEER数据库中壶腹周围癌病人临床数据,根据设定标准最终纳入1775例,按7∶3的比例分为建模组(1242例)和验证组(533例)。在建模组中,通过Cox比例风险回归模型筛选影响壶腹周围癌预后的危险因素,在回归分析结果的基础上构建预后预测模型并绘制Nomogram图。分别在建模组和验证组中对模型的预测效能进行验证。结果Cox回归模型多因素分析结果显示年龄、T分期、N分期、肿瘤病理分级、肿瘤病理类型、是否手术为壶腹周围癌的危险因素,将上述6个变量纳入预测模型,绘制Nomogram图,进行1年、3年、5年生存率预测。在建模组和验证组中C指数分别为0.7047(95%CI:0.6854,0.7241)和0.7001(95%CI:0.6689,0.7314)。建模组中1年、3年、5年的ROC曲线的AUC值分别为0.766、0.756、0.757,验证组中1年、3年、5年的ROC曲线的AUC值分别为0.736、0.733和0.742。校正曲线显示,预测生存率与实际生存率有较好的一致性,决策曲线显示预测模型具有一定的临床价值。结论构建的壶腹周围癌预后预测模型具有良好的预测价值。

Objective To analyze the risk factors for the prognosis of Periampullary carcinoma and establish a prognostic model.Methods Clinical data of periampulla carcinoma patients from SEER database were retrospectively analyzed.According to the set criteria,1775 patients were included and divided into the modeling group(1242 case)and the validation group(533 case)in a 7∶3 ratio.In the modeling group,Cox proportional risk regression model was used to screen the risk factors influencing the survival and prognosis of periampullary carcinoma,and a Nomogram was constructed based on the regression analysis results.The predictied efficiency of the model was verified in the modeling group and the verification group respectively.Results Age,T stage,N stage,degree of tumor differentiation,pathological type and operation were found to be risk factors for periampullary carcinoma by Cox regression model.The above six variables were included in the prediction model and a Nomogram was drawn to predict 1year,3year and 5year survival rates.C index in modeling group and verification group was 0.7047(95%CI:0.6854,0.7241)and 0.7001(95%CI:0.6689,0.7314),respectively.The AUC values of 1year,3year and 5year ROC curves in the modeling group were 0.766、0.756 and 0.757,respectively,and those of 1year,3year and 5year ROC curves in the verification group were 0.736、0.733 and 0.742,respectively.The correction curve showed that the predictors of survival rate was consistent with the actual survival rate,and the decision curve showed that the prediction model had certain clinical value.Conclusion The prognostic model of periampullary carcinoma has good predictive value.