未转移结直肠癌患者术后转移风险预测列线图的构建

摘要

目的 构建结直肠(colorectal cancer,CRC)患者预后列线图预测模型,并验证其准确性.方法 回顾性分析就诊于苏州大学第三附属医院的438例CRC患者的临床病理资料.建立COX单因素、多因素回归模型确定预后的独立危险因素.用R软件建立列线图,绘制术后3年、5年无病生存率(disease free survival,DFS)校准图线并与实际观察情况比较.用Bootstrap法进行内部验证,计算一致性指数(C-index)评估模型准确性.分析时间依赖的ROC曲线比较其与第7版美国癌症联合委员会(American Joint Committee On Cancer,AJCC)分期系统(TNM系统)在预测术后的3年、5年的DFS的敏感性和特异性.结果 438例CRC患者中233例患者出现转移,其中肝脏转移105例,肺转移57例.COX回归模型分析示肿瘤分化程度,肿瘤TNM分期,癌胚抗原(CEA)水平,糖类抗原19-9(CA19-9)水平,中性粒细胞与淋巴细胞之比(Neutrophil lymphocyte radio,NLR),抑癌基因P53是患者预后的独立影响因素.列线图用于预测生存的C指数为0.678.校正曲线表明预测的3年、5年DFS与实际观察情况高度符合.时间依赖的ROC曲线结果表明,相较于传统AJCC-TNM分期,列线图预后模型在预测术后3年和5年的DFS具有更高的敏感性和特异性.结论 列线图可较准确预测个体CRC患者的预后,利于临床工作者对其随访或及时开展对患者有益治疗.%Objective To construct a prognostic nomogram for predicting the prognosis of patients with colorectal cancer ( CRC) , and verify its accuracy. Methods The clinical pathologic data from 438 CRC patients hospitalized in the Third Affiliated Hospital of Soo-chow University during January 2006 and May 2013 were retrospectively analyzed. The independent risk factors for predicting the prog-nosis of CRC were determined by the univariate and multivariate regression model. The prognostic nomogram was established by the R-language software. Then, the nomograms of postoperative 3-year and 5-year disease free survivals ( DFS) were drawn, and compared with the actual status. The internal validation and accuracy of the nomogram were determined by the Bootstrap method and the calculat-ed concordance index ( C-index) , respectively. The sensitivity and specificity of the nomogram for predicting the 3-year and 5-year DFS were compared with those of TNM system established by the American Joint Committee On Cancer (AJCC) (7th ed.) by using the time-dependent ROC curve. Results Among 438 CRC patients, the metastasis of CRC occurred in 233 patients, including 105 liver metas-tasis and 57 lung metastasis. Multivariate COX regression analysis showed that tumor differentiation degree, TNM stage, serum CEA level, serum CA19-9 level, neutrophil lymphocyte ratio ( NLR) and P53 level were the independent risk factors of CRC. The C-index of the constructed nomogram for predicting the survival rate of CRC patients was 0.678. The predicted 3-year and 5-year DFS by the no-mogram were highly coincident with the actual status. The analysis results of the time-dependent ROC curve showed that the sensitivity and specificity of the established nomogram for predicting the postoperative 3-year and 5-year DFS were higher than those of AJCC-TNM stage.Conclusion The established nomogram may accurately predict the prognosis of CRC patients, which may be helpful for clinicians to follow up or make beneficial treatment for CRC patients.