摘要
目的探讨Ⅱ~Ⅲ期结肠癌根治术后复发危险因素及其列线图预测模型的应用价值。方法采用回顾性病例对照研究方法。收集2013年1月至2016年6月西安交通大学第一附属医院收治的228例行根治性切除术治疗Ⅱ~Ⅲ期结肠癌病人的临床病理资料;男118例,女110例;中位年龄为62岁,年龄范围为25~87岁。所有病人行开腹或腹腔镜辅助结肠癌根治性切除术。观察指标:(1)术后复发情况。(2)影响Ⅱ~Ⅲ期结肠癌根治术后复发的危险因素分析。(3)Ⅱ~Ⅲ期结肠癌根治术后复发列线图预测模型的构建及评价。采用门诊或电话方式进行随访,了解病人术后3年复发情况。随访时间截至2019年6月。偏态分布的计量资料以M(范围)表示。计数资料以绝对数表示,组间比较采用Pearsonχ2检验或Fisher确切概率法。多因素分析采用Logistic逐步回归分析。将独立危险因素引入R 3.6.1软件,构建列线图预测模型。绘制受试者工作特征曲线(ROC),以曲线下面积(AUC)评价列线图预测模型的区分度。使用R软件绘制校准度曲线图评价列线图预测模型的一致性。结果(1)术后复发情况:228例病人中,53例术后复发,其中局部复发19例,远处转移34例。34例远处转移病人中,肝转移14例、肺转移7例、脑转移4例、多发转移及其他部位单发转移9例。53例病人术后复发时间为12个月(6~19个月)。(2)影响Ⅱ~Ⅲ期结肠癌根治术后复发的危险因素分析:单因素分析结果为肠梗阻、术前癌胚抗原(CEA)、腹腔积液、血管侵犯是影响Ⅱ~Ⅲ期结肠癌根治术后复发的相关因素(χ^(2)=4.463、13.622、10.914、5.911,P<0.05)。病理学N分期是影响Ⅱ~Ⅲ期结肠癌根治术后复发的相关因素(P<0.05)。多因素分析结果显示:术前CEA>5μg/L、腹腔积液、血管侵犯、病理学N分期为N1期或N2期是影响Ⅱ~Ⅲ期结肠癌根治术后复发的独立危险因素(优势比=3.129,3.071,7.634,3.439,15.467,95%可信区间为1.328~7.373,1.047~9.007,1.103~52.824,1.422~8.319,3.498~68.397,P<0.05)。(3)Ⅱ~Ⅲ期结肠癌根治术后复发列线图预测模型的构建及评价:根据多因素分析结果,将术前CEA、腹腔积液、血管侵犯及病理学N分期引入R 3.6.1软件,构建Ⅱ~Ⅲ期结肠癌根治术后复发的列线图预测模型。术前CEA>5μg/L的列线图评分为41.7分,腹腔积液为41.0分,血管侵犯为74.2分,病理学N分期N1期为45.1分、N2期为100.0分,各项危险因素不同取值得分总和对应术后复发概率。绘制ROC评价列线图预测Ⅱ~Ⅲ期结肠癌根治术后复发的能力,其AUC为0.805(95%可信区间为0.737~0.873,P<0.05)。校准曲线图显示Ⅱ~Ⅲ期结肠癌根治术后列线图模型预测复发概率与实际复发概率具有较好一致性。结论术前CEA>5μg/L、腹腔积液、血管侵犯、病理学N分期为N1或N2期是Ⅱ~Ⅲ期结肠癌根治术后复发的独立危险因素;以此构建列线图预测模型有助于预测Ⅱ~Ⅲ期结肠癌根治术后复发风险。
Objective To investigate the risk factors for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer,and application value of a nomogram prediction model.Methods The retrospective case‐control study was conducted.The clinicopathological data of 228 patients with stageⅡ-Ⅲcolon cancer who underwent radical resection in the First Affiliated Hospital of Xi′an Jiaotong University from January 2013 to June 2016 were collected.There were 118 males and 110 females,aged from 25 to 87 years,with a median age of 62 years.All patients underwent open or laparoscopic‐assisted radical resection of colon cancer.Observation indicators:(1)postoperative tumor recurrence;(2)risk factors analysis for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer;(3)development and evaluation of a nomogram prediction model for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer.Follow‐up using outpatient examination and telephone interview was performed to detect postoperative 3‐year tumor recurrence up to June 2019.Measurement data with skewed distribution were represented M(range).Count data were described as absolute numbers,and comparison between group analyzed using the Pearson chi‐square test or Fisher exact probability.Multivariate analysis was performed using Logistic stepwise regression analysis.The independent risk factors were included into R 3.6.1 software to construct a nomogram prediction model.The receiver operating characteristic curve(ROC)was drawed,and the area under curve(AUC)was used to evaluate discrimination of the nomogram prediction model.The calibration chart with R software was used to evaluate consistency of the nomogram prediction model.Results(1)Postoperative tumor recurrence:53 of 228 patients had postoperative tumor recurrence including 19 cases with locoregional recurrence and 34 cases with distant metastasis.Of the 34 patients with distant metastasis,there were 14 cases with liver metastasis,7 cases with lung metastasis,4 cases with brain metastasis,and 9 cases with multiple metastasis or isolated metastasis in other sites.The time to recurrence was 12 months(range,6-19 months).(2)Risk factors analysis for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer:results of univariate analysis showed that bowel obstruction,preoperative carcinoembryonic antigen(CEA)level,ascites,vascular invasion were related factors for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer(χ^(2)=4.463,13.622,10.914,5.911,P<0.05).Pathological N stage was also a related factor for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer(P<0.05).Results of multivariate analysis showed that preoperative CEA level>5μg/L,ascites,vascular invasion and pathological N stage as stage N1 or N2 were independent risk factors for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer(odds ratio=3.129,3.071,7.634,3.439,15.467,95%confidence interval as 1.328-7.373,1.047-9.007,1.103-52.824,1.422-8.319,3.498-68.397,P<0.05).(3)Development and evaluation of a nomogram prediction model for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer:based on preoperative CEA level,ascites,vascular invasion and pathological N stage multivariate analysis,a nomogram prediction model for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer was developed using R 3.6.1 software.The nomogram score was 41.7 preoperative CEA level>5μg/L,41.0 for ascites,74.2 for vascular invasion,45.1 and 100.0 for pathological N stage as stage N1 and N2,respectively.The total of different scores for risk factors corresponded to the probability of postoperative recurrence.The ROC of nomogram for recurrence after radical resection of stageⅡ-Ⅲcolon cancer was drawed,with the AUC of 0.805(95%confidence interval as 0.737-0.873,P<0.05).The calibration chart showed a good consistency between the probability of recurrence after radical resection of stageⅡ-Ⅲcolon cancer predicted by nomogram and the actual probability of postoperative recurrence.Conclusions Preoperative CEA level>5μg/L,ascites,vascular invasion and pathological N stage as stage N1 or N2 are independent risk factors for tumor recurrence after radical resection of stageⅡ-Ⅲcolon cancer.The nomogram prediction model contributes to prediction of the recurrent risks after radical resection of stageⅡ-Ⅲcolon cancer.
作者
程晨
吴云桦
徐正水
赵晨野
李小鹏
余钧辉
国婧
郑见宝
魏光兵
孙学军
Cheng Chen;Wu Yunhua;Xu Zhengshui;Zhao Chenye;Li Xiaopeng;Yu Junhui;Guo Jing;Zheng Jianbao;Wei Guangbing;Sun Xuejun(Department of General Surgery,the First Affiliated Hospital of Xi'an Jiaotong University,Xi′an 710061,China)
出处
《中华消化外科杂志》
CAS
CSCD
北大核心
2021年第3期331-338,共8页
Chinese Journal of Digestive Surgery
基金
国家自然科学基金(81972720)。
关键词
结肠肿瘤
根治术
复发
列线图
预测模型
危险因素
Colonic neoplasms
Radical resection
Recurrence
Nomogram
Prediction model
Risk factors