基于术前临床及超声资料的列线图模型预测卵巢癌患者淋巴结转移的临床价值

Clinical value of nomogram model based on preoperative clinical and ultrasonic data in predicting lymph node metastasis in ovarian cancer patients

目的 基于术前临床及超声资料建立列线图模型,探讨其预测卵巢癌患者淋巴结转移的临床价值。方法 选取我院经手术病理确诊的卵巢癌患者475例,按照7∶3的比例将其随机分为训练集333例和验证集142例,其中训练集中发生淋巴结转移92例(LNM组),未发生淋巴结转移241例(NLNM组)。比较两组临床及超声资料的差异。应用Logistic回归分析筛选预测卵巢癌患者淋巴结转移的独立影响因素,并绘制列线图。绘制受试者工作特征(ROC)曲线,采用曲线下面积和一致性指数(C-index)评价列线图模型的区分度;校准曲线和Brier评分评价其校准度;决策曲线评价其临床适用性。结果 两组肿瘤组织学分型、国际妇产科联盟(FIGO)分期及位置比较差异均有统计学意义(均P<0.01);且NLNM组Ki-67表达水平、多灶性占比、最大径均小于LNM组,差异均有统计学意义(均P<0.001)。肿瘤最大径、多灶性及Ki-67表达水平均为预测卵巢癌患者淋巴结转移的独立危险因素(OR=1.010、5.236、1.023,均P<0.01)。ROC曲线分析显示,列线图模型预测训练集、验证集中卵巢癌患者淋巴结转移的曲线下面积分别为0.819(95%可信区间:0.770~0.868)、0.794(95%可信区间:0.717~0.870),C-index分别为0.837(95%可信区间:0.811~0.894)、0.749(95%可信区间:0.712~0.791),提示模型具有良好的区分度。校准曲线分析显示,列线图模型在训练集和验证集中的预测曲线与校准曲线均较为贴合,Brier评分分别为0.22、0.38,提示模型预测概率与实际概率的一致性较高,校准度较好。决策曲线分析显示,列线图预测训练集、验证集中卵巢癌患者淋巴结转移的临床适用性均较好。结论 基于术前临床及超声资料的列线图模型在预测卵巢癌患者淋巴结转移中有一定的临床价值。

Objective To establish a nomogram model based on preoperative clinical and ultrasonic data,and to explore its clinical value in predicting lymph node metastasis in ovarian cancer patients.Methods A total of 475 ovarian cancer patients confirmed by surgical pathology in our hospital were selected.They were randomly divided into a training set(n=333)and a validation set(n=142)at a ratio of 7∶3.There were 92 cases of lymph node metastasis(LNM group)and 241 cases without lymph node metastasis(NLNM group)in the training set.The clinical and ultrasonic data of LNM group and NLNM group were compared.Logistic regression analysis was applied to screen the independent influencing factors for predicting lymph node metastasis in ovarian cancer patients,and the nomogram was drawn.Receiver operating characteristic(ROC)curve was drawn,the area under the curve and consistency index(C-index)were used to evaluate the differentiation of the nomogram.Calibration curve and Brier score were used to evaluate the calibration degree.Decision curve was used to evaluate the clinical applicability.Results There were statistically significant differences in histological types,FIGO stage and tumor location between the two groups(all P<0.01).Additionally,the expression level of Ki-67,proportion of multifocality and maximum diameter in the NLNM group were lower than those in the LNM group,with statistically significant differences(all P<0.001).Tumor maximum diameter,multifocality and the expression level of Ki-67 were independent risk factors for predicting lymph node metastasis in ovarian cancer patients(OR=1.010,5.236,1.023,all P<0.01).ROC curve analysis showed that the area under the curve of the nomogram model for predicting lymph node metastasis in ovarian cancer patients in the training and validation sets were 0.819(95%CI:0.770~0.868)and 0.794(95%CI:0.717~0.870),and the C-index were 0.837(95%CI:0.811~0.894),0.749(95%CI:0.712~0.791),respectively,indicating good differentiation of the model.Calibration curve analysis showed that the calibration degree of the prediction curve and calibration curve judged by the nomogram model was higher in the training and the validation sets,Brier score were 0.22,0.38,indicating good consistency between predicted and actual outcomes,and the calibration degree was good.Decision curve analysis showed good clinical applicability of the nomogram model for predicting lymph node metastasis in ovarian cancer patients in the training and the validation sets.Conclusion The nomogram model based on preoperative clinical and ultrasonic data has a certain clinical value in predicting lymph node metastasis in ovarian cancer patients.