基于列线图模型预测结直肠癌淋巴结转移率与临床特征及预后的关系

Correlation of metastatic lymph node ratio with clinical characteristics and prognosis in colorectal cancer predicted by nomogram model

目的探讨列线图模型预测结直肠癌淋巴结转移率(LNR)与临床特征及预后的关系。方法收集260例结直肠癌患者的临床特征及预后信息,采用受试者工作特征(ROC)曲线计算最佳截断值并分组,通过Kaplan-Meier绘制生存曲线,Cox回归模型筛选出与预后相关的因素,采用R软件构建列线图模型进行可视化分析。结果LNR预测结直肠癌患者预后的最佳截断值为0.12,根据最佳截断值将260例结直肠癌患者分为LNR低值组(n=172,LNR﹤0.12)及LNR高值组(n=88,LNR≥0.12)。LNR低值组与LNR高值组结直肠癌患者TNM分期、分化程度、pT分期、pN分期比较,差异均有统计学意义(P﹤0.05)。LNR高值组患者中位无病生存期(DFS)和中位总生存期(OS)均明显短于LNR低值组,差异均有统计学意义(P﹤0.01)。Cox回归分析结果显示,TNM分期为Ⅲ~Ⅳ期和LNR高值均为结直肠癌患者DFS、OS的危险因素(P﹤0.01)。依此建立列线图预测模型显示,随着TNM分期与LNR增加,列线图模型相应评分增高,相对应的3年及5年复发及死亡风险上升。DFS列线图模型的预测能力一致性指数(C-index)为0.689(95%CI:0.642~0.736),OS列线图模型的预测能力C-index为0.666(95%CI:0.607~0.725)。结论基于LNR及TNM分期建立的预测结直肠癌患者复发及死亡风险的列线图模型具有良好的临床应用价值。

Objective To investigate the correlation of metastatic lymph node ratio(LNR)with clinical characteris-tics and prognosis in colorectal cancer(CRC)predicted by nomogram model.Method Clinical characteristics and prog-nosis of 260 CRC patients were collected.The optimal cut-off value was determined by the receiver operating characteris-tic(ROC)curve.Survival curve was plotted by the Kaplan-Meier method,Cox regression model was applied to screen out prognostic factors and R software was used to construct a nomogram model for visualization analysis.Result The optimal cut-off value of LNR in predicting prognosis of CRC patients was 0.12,which divided 260 CRC patients into low LNR group(n=172,LNR<0.12)and high LNR group(n=88,LNR≥0.12).There were significant differences for TNM stage,differentiation degree,pT stage and pN stage between low LNR group and high LNR group(P<0.05).The median disease-free survival(DFS)and median overall survival(OS)in high LNR group were shorter than those in low LNR group(P<0.01).Cox regression analysis showed that TNM stage III-IV and high LNR were risk factors for DFS and OS in CRC patients(P<0.01).According to the established nomogram prediction model,with the increase of TNM stage and LNR,the corresponding scores of nomogram model increased,and the corresponding 3-year and 5-year recurrence and death risk increased.The C-index of prediction ability of DFS and OS nomogram model were 0.689(95%CI:0.642-0.736)and 0.666(95%CI:0.607-0.725)respectively.Conclusion The nomogram model based on LNR and TNM stages to predict recurrence and death risk of CRC patients has good clinical application value.