基于临床病理特征构建胃癌根治术后早期复发风险预测模型

A clinicopathological characteristics based model to predict early recurrence of patients with gastric cancer after radical surgery

目的构建胃癌患者根治术后早期复发风险预测模型,指导个体化治疗与随访。方法收集接受根治性手术治疗的胃癌患者114例,通过最小绝对收缩和选择算子回归分析进行变量筛选,应用COX回归分析构建预测模型并绘制列线图,通过受试者工作特征曲线(ROC)、log-rank检验、COX回归分析和校准图评价其预测的准确性。结果通过变量筛选与COX回归分析,构建了基于年龄、性别、黏液腺癌、中分化腺癌、Lauren混合型、病理N分期以及病理分期7个临床病理特征参数的复发指数(RI)预测模型,RI的1年、2年ROC曲线下面积> 0. 7,是术后无复发生存的独立风险因素,根据RI,低复发风险患者RFS明显长于高复发风险患者。列线图预测的术后1年、2年复发风险与理想预测值较接近。结论该模型可以较准确地预测胃癌患者术后早期复发风险,值得在临床上进一步证实与推广。

Objective To construct a predictive model of early recurrence for gastric patients after radical gastrectomy based on clinicopathological characteristics and guide individualized adjuvant therapy and follow-up.Methods A total of 114 patients with gastric cancer who underwent radical surgery were enrolled.Least absolute shrinkage and selection operator regression analysis was used to screen for factors associated with recurrence-free survival(RFS)and a predictive model was constructed based on the factors through COX regression analysis.The model was validated by receiver operating characteristic curve,log-rank test,univariate and multivariate analyses.In addition,a nomogram based on the model was generated and evaluated by calibration curves.Results A predictive model was developed and recurrence indexes(RI=0.019×age+0.438×male+(-0.813)×mucinous adenocarcinoma+(-0.446)×moderately differentiated adenocarcinoma+0.30×Lauren mixed type+0.472×lymphovascular or neural invasion+0.224×pathological N stage+0.168×pathological stage)were calculated.The areas under the ROC curves of RI to predict 1-year and 2-year recurrence were greater than 0.7.The patients with low RI had a longer RFS.The RI was also an independent marker to predictor early recurrence.The nomogram can predict probabilities of 1-year and 2-year recurrence accurately.Conclusions This model can accurately predict the risk of early recurrence after radical surgery in patients with gastric cancer,which is worthy of further confirmation and wide application in clinical practice.