基于3.0T MR增强影像构建鉴别结直肠癌周围癌结节与转移性淋巴结的预测模型  

Construction of a predictive model based on 3.0T MRI-enhanced imaging for differentiating tumor depo- sits from metastatic lymph nodes in patients with colorectal cancer

作　　者：周之怀 王龙胜[1] 刘三春[2] 张玉 朱越 ZHOUZhi-huai;WANG Long-sheng;LIU San-chun(Department of Radiology,the Second AffiliatedHospital of Anhui Medical University,Hefei 230601,China)

机构地区：[1]安徽医科大学第二附属医院放射科,安徽合肥230601 [2]蚌埠医学院第二附属医院放射科,233040

出　　处：《放射学实践》2023年第12期1562-1567,共6页Radiologic Practice

基　　金：临床医学学科建设项目(2021lcxk029);安徽省质量工程项目(2019jyxm1005);蚌埠医学院自然科学类项目(BYKY2019175ZD)。

摘　　要：目的:探讨基于3.0T MR增强影像构建的预测模型对结直肠癌周围癌结节(TD)与转移性淋巴结(MLN)的鉴别诊断价值。方法:分析经手术病理确诊为结直肠癌伴有灶周TD和/或MLN的106例患者的病例资料。其中,A组22例(20.8%),共发现28个灶周TD和32个MLN;B组84例(79.2%),仅发现150个MLN。所有患者在术前1周内行MRI检查,扫描序列主要包括T1WI、FS-T2WI、DWI及肝脏快速容积增强成像(LAVA+)。主要评估征象:结节数目、形态、短径、ADC值、强化方式及强化率。两组之间临床资料及灶周结节(TD/MLN)MRI特征中连续变量的比较采用独立样本t检验,分类变量的比较采用χ2检验或Fisher精确检验,等级变量采用Wilcoxon符号秩和检验。对组间差异有统计学意义(P<0.05)的变量进行二元Logistic回归分析,建立预测模型,并采用ROC曲线评估模型对TD与MLN的鉴别效能。结果:两组患者的临床和病理资料中,性别、年龄、肿瘤发生部位、血清肿瘤标记物CA19-9和CA50水平、肿瘤分化程度、周围脉管及神经侵犯情况的差异均无统计学意义(P>0.05);而肿瘤周围结节的数目及血清肿瘤标记物CEA的水平在两组之间的差异均具有统计学意义(P<0.05)。两组灶周结节MRI征象的比较:强化方式的差异无统计学意义(P>0.05),而结节形态、短径、ADC值及强化率的差异均具有统计学意义(P<0.05)。二元Logistic回归分析发现结节强化率和短径是预测TD的独立影响因素(P<0.05)。ROC曲线分析显示当结节短径>10.45 mm时,诊断TD的AUC、敏感度和特异度分别为0.719、67.7%和73.6%;当结节强化率<75.0%,诊断TD的AUC、敏感度和特异度分别为0.869、83.2%和80.1%。构建的影像预测模型的AUC为0.872,高于结节强化率和短径的AUC,且其诊断符合率、敏感度和特异度分别为89.7%、84.6%和73.5%。结论:结节短径及强化率是鉴别结直肠癌周围TD与MLN的独立影响因素,且两者构建的影像预测模型Objective:To explore the value of constructing a predictive model based on 3.0T MRI-enhanced imaging for differentiating tumor deposits(TDs)from metastatic lymph nodes(MLNs)in patients with colorectal cancer.Methods:A analysis was conducted on data from 106 cases of pathologically confirmed colorectal cancer with perifocal nodules(TDs and/or MLNs).Among them,22 patients(20.8%)had TDs,with a total of 28 TDs and 32 MLNs,and 84 patients(79.2%)had a total of 150 MLNs.All patients underwent MRI examination within one week before surgery,with scanning sequences mainly including T 1WI,FS-T 2WI,DWI,and liver acceleration volume acquisition enhancement scan(LAVA+).MRI image evaluation mainly included the number,morphology,short diameter,apparent diffusion coefficient(ADC),enhancement mode,and enhancement rate of the nodules.The clinical data and MRI features of perifocal nodules were compared statistically between TD group and MLN group,independent sample t-test was used for continuous variables,χ2-test or Fisher exact test was used for categorical variables,and Wilcoxon rank-sum test was used for ordinal variables.Variables with statistical significance(P<0.05)were subjected to binary logistic regression analysis,and a predictive model was established by drawing ROC curve to evaluate the discrimination efficacy between TD and MLN.Results:There were no statistically significant difference in gender,age,tumor location,tumor markers(CA19-9,CA50),differentiation degree,or perivascular and perineural invasion between the two groups(P>0.05),while the number of nodules around the tumor and CEA level had statistical significance(P<0.05).In the MRI signs,there was no statistically significant difference in the enhancement mode of perifocal nodules between the two groups(P>0.05),while the morphology,short diameter,ADC value,and enhancement rate of the nodules all had statistical significance(P<0.05).Binary logistic regression analysis revealed that the enhancement rate and short diameter of the nodules were independent influencin

关 键 词：结直肠肿瘤 癌结节 转移性淋巴结 表观扩散系数 强化率 预测模型 

分 类 号：R445.2[医药卫生—影像医学与核医学] R735.5[医药卫生—诊断学]