胰腺癌患者术后生存预测模型的构建

Construction of a postoperative survival prediction model for pancreatic cancer patients

目的:探索胰腺癌患者的预后因素,构建胰腺癌患者术后生存预测模型并予以优化。方法:选取弋矶山医院2014~2020年确诊为胰腺导管腺癌(PDAC)并采取手术治疗的118例患者作为研究对象,统计患者的临床病理特征及其他相关资料,使用SPSS 26.0统计软件进行单因素、多因素Cox比例风险回归分析(Kaplan-Meier法)相关影响因素,使用r 4.1.1软件绘制胰腺癌术后预测的列线图模型、受试者工作特征(ROC)曲线、校准曲线及决策曲线。结果:118名PDAC患者随访1~84个月,中位生存时间17个月,手术后1、2、3年的生存率分别为62.60%、33.70%、20.90%,多因素COX比例风险回归分析显示有化疗是胰腺癌患者术后影响的保护因素(P<0.05),低分化程度、有神经侵犯、有胰漏、TNM分期为Ⅲ期及ALBI和PI评分高是胰腺癌患者术后影响的危险因素(P<0.05),将上述变量纳入列线图模型,经内部验证后,该nomogram模型的C-index为0.861,术后第1~3年的AUC均较高(>0.85),校准曲线贴合良好,决策曲线表明模型的预测效能较高。结论:本研究构建的胰腺癌患者术后生存预测模型能够有效预测胰腺癌患者术后1~3年生存率,并能根据患者的预后影响因素提供针对性的治疗。

Objective:To investigate the prognostic factors of pancreatic cancer patients,construct and optimize a postoperative survival prediction model for these patients.Methods:The study included 118 patients,confirmed as pancreatic ductal adenocarcinoma(PDAC)and treated surgically between 2014 and 2020 in our hospital.The clinicopathological characteristics and other relevant data of patients were analyzed,using univariate and multifactor Cox proportional risk regression analyses by SPSS 26.0.Kaplan Meier method was used for survival analysis.Factors affecting pancreatic cancer prognosis were identified according to Cox proportional risk regression analysis.r 4.1.1 software was used to draw nomograph model,receiver operating characteristic(ROC)curve,calibration curve,and decision curve.r esults:A total of 118 PDAC patients participated in the study,with a follow-up time range of 1 to 84 months and a median survival time of 17 months.The 1-year,2-year,and 3-year survival rates after surgery were 62.60%,33.70%,and 20.90%respectively.Multivariate COX proportional risk regression analysis showed that chemotherapy was a protective factor for prognosis in pancreatic cancer patients(P<0.05),and the risk factors were involved in low differentiation,nerve invasion,pancreatic leakage,tumor node metastasis(TNM)stageⅢ,and high albumin-bilirubin(ALBI)and predictive index(PI)scores(P<0.05).The above variables were included in the nomogram model.The nomogram was validated internally,and the consistency index(C-index)was 0.861.The areas under the curve(AUC)at 1 st,2 nd and 3 rd year were high(>0.85),the calibration curve fitted well,and the decision curve analysis indicated the model had favorable predictive efficiency.Conclusion:The newly constructed postoperative survival prediction model for pancreatic cancer patients can effectively predict 1-year,2-year and 3-year survival rates and provide targeted treatment based on factors affecting patient′s prognosis.