乳腺派吉特病预后风险列线图模型构建及验证评价

Construction and validation evaluation of prognostic risk nomogram model of Paget disease of breast

目的探讨影响乳腺Paget病(MPD)患者预后风险因素并建立预后风险列线图模型。方法通过SEER数据库筛选2004-2014年期间MPD患者数据,采用单因素分析其中影响MPD患者预后相关因素,通过多因素Cox比例风险回归分析以筛选影响其1,3,5年生存率的独立风险因素构建模型,利用一致性指数(C指数)评估其预测价值并用1000次Bootstrap法检验其预测效能;后利用校准曲线对模型预测性能进行验证,受试者工作特征曲线(ROC)及决策曲线(DCA)分别评估模型较AJCC分期系统的敏感性及实用性。结果该模型的C指数为0.795(95%置信区间(CI)0.773~0.818)。校准曲线表现出良好一致性;ROC曲线显示了模型预测MPS 1、3、5年生存率的敏感性;DCA证实了模型预测MPD患者1、3、5年生存率优于AJCC分级。结论年龄、婚姻状态、AJCC分级、手术、放疗、化疗和婚姻状态转换MPD患者的独立预后因素,依此建立的列线图模型较传统的AJCC分级具有更高的精准度和临床价值。

Objective To explore the prognostic risk factors of patients with breast Paget disease(PD)and establish a prognostic risk nomogram model.Methods The data of PD patients from 2004 to 2014 were screened through the SEER database.The factors affecting the prognosis of PD patients were analyzed by univariate analysis and the independent risk factors affecting the 1,3,and 5 year survival rates were screened by multivariate Cox proportional hazard regression analysis coefficient.The model was constructed with independent risk factors,the consistency index(C index)was used to evaluate the predictive value and the predictive efficiency was tested by 1000 times bootstrap method,then the calibration curve was used to verify the predictive performance of the model,the receiver operating characteristic curve(ROC)and the decision curve(DCA)were separately used to assess the sensitivity and practicality of the model compared to the AJCC staging system.Results The C index of the model was 0.795(95%confidence interval(CI)0.773~0.818).The calibration curve showed good consistency,the ROC curve showed the sensitivity of the model to predict the survival rate of MPD 1,3,and 5 years.DCA confirmed that the model predicted that the 1,3,and 5-year survival rate of PD patients was better than that of AJCC.Conclusion Age,marital status,AJCC classification,surgery,radiotherapy,chemotherapy and marital status conversion are independent prognostic factors for PD patients.The nomogram model established based on this has higher accuracy and clinical value than the traditional AJCC classification.