原发性肝癌发生血管侵犯的危险因素分析

Risk factors for vascular invasion of primary liver cancer

目的探讨影响原发性肝癌血管侵犯的危险因素。方法收集2013年1月-2014年6月于解放军第三○二医院住院的211例肝癌患者的临床资料,其中133例用于logistic回归模型拟合,78例用于该模型的检验。原发性肝癌血管侵犯影响因素分析采用单因素和多因素logistic回归分析,并建立回归模型,绘制受试者工作特征曲线(ROC曲线),确定预测血管侵犯模型的最佳临界值。结果单因素logistic回归分析显示,肿瘤直径、中性粒细胞/淋巴细胞比值(NLR)、血小板/淋巴细胞比值(PLR)、纤维蛋白原(Fb)和淋巴结转移为肝癌血管侵犯的危险因素(95%可信区间:1.376~1.846、1.847~4.195、1.008~1.024、1.608~3.274,P值均=0);经多因素logistic回归分析筛选出肿瘤直径、PLR、NLR、Fb(95%可信区间:1.250~1.815、0.173~1.998、1.411~4.397、1.008~2.193,P值分别为0、0.002、0.022、0.046)等4个因素用于回归模型的拟合。ROC曲线下面积为0.927,95%CI:0.881~0.973,ROC曲线显示最大灵敏度、特异度分别为92.9%、86.5%,预测准确率为82.79%。结论采用肿瘤直径、PLR、NLR、Fb代入模型拟合回归方程的预测准确率较高,对原发性肝癌血管侵犯的早期预警有一定的参考价值。

Objective To investigate the risk factors for vascular invasion of primary liver cancer （PLC）.Methods A retrospective analysis was performed for the clinical data of 211 patients with liver cancer who were hospitalized in 302 Hospital of PLA from January 2013 to June 2014.The logistic regression model fitting was used for the data of 133 patients,and the data of the other 78 patients was used for the verification of this model.Univariate and multivariate logistic regression analyses were used to identify the influencing factors for vascular invasion of PLC;the logistic regression model was established and the receiver operating characteristic （ROC） curve was plotted to determine the optimal cut-off value of this model.Results The univariate logistic regression analysis showed that tumor diameter （odds ratio [OR]=1.594,95％ confidence interval [CI]：1.376-1.846,P =0）,neutrophil-lymphocyte ratio （NLR） （OR =2.783,95％ CI：1.847-4.195,P=0）,platelet-to-lymphocyte ratio （PLR） （OR=1.016,95％ CI：1.008-1.024,P=0）,fibrinogen （Fb） （OR=2.295,95 ％ CI：1.608-3.274,P =0）,and lymph node metastasis （OR =11.664,95 ％ CI：3.744-36.338,P =0） were risk factors for vascular invasion of PLC;the multivariate logistic regression analysis showed that tumor diameter （OR =1.506,95％ CI：1.250-1.815,P =0）,PLR （OR=1.499,95％ CI：0.173-0.998,P=0.022）,NLR （OR =2.491,95％ CI：1.411-4.397,P=0.002）,and Fb（OR =1.486,95％ CI：1.008-2.193,P =0.046） were used for regression model fitting.The area under the ROC curve was 0.927 （95％ CI：0.881-0.973）,and the ROC curve showed that the model had the highest sensitivity of 92.9％,the highest specificity of 86.5％,and a prediction accuracy rate of 82.79％.Conclusion The regression equation containing tumor diameter,PLR,NLR,and Fb established in this study has a high prediction accuracy of vascular invasion of PLC and provides a reference for early warning of vascular invasion of PLC.