基于FIB、HB、FOBT、CEA构建的Logistic回归模型对结直肠癌的诊断价值  

Diagnostic value of constructed Logistic regression model in colorectal cancer based on FIB,HB,CEA,FOBT and CEA

作　　者：侯晓慧 张玉英 谢小亮[3] 王良方 冯丽君 杜沛静 李刚 师志云 HOU Xiaohui;ZHANG Yuying;XIE Xiaoliang;WANG Liangfang;FENG Lijun;DU Peijing;LI Gang;SHI Zhiyun(Department of Medical Experimental Center,General Hospital of Ningxia Medical University,Yinchuan,Ningxia 750000,China;First Clinical Medical College,Ningxia Medical University,Yinchuan,Ningxia 750000,China;Department of Colorectal Surgery,General Hospital of Ningxia Medical University,Yinchuan,Ningxia 750000,China)

机构地区：[1]宁夏医科大学总医院医学实验中心,宁夏银川750000 [2]宁夏医科大学第一临床医学院,宁夏银川750000 [3]宁夏医科大学总医院结直肠外科,宁夏银川750000

出　　处：《检验医学与临床》2024年第6期805-809,共5页Laboratory Medicine and Clinic

基　　金：宁夏自然科学基金项目(2023AAC03516);宁夏医科大学总医院“青年拔尖人才”专项(NYDZY026)。

摘　　要：目的 评价纤维蛋白原(FIB)、血红蛋白(HB)、癌胚抗原(CEA)和粪便隐血试验(FOBT)联合检测在结直肠癌患者中的诊断价值,为结直肠癌诊断提供参考。方法 选取宁夏医科大学总医院2021年1月至2022年6月结直肠外科收治的110例结直肠癌患者作为癌症组,选取同期病房收治的43例良性结直肠疾病患者为非癌症组。回顾性分析并比较癌症组及非癌症组常规实验室指标(血常规、生化常规、肿瘤标志物、细胞因子、FOBT)结果,采用Logistic回归分析结直肠癌的影响因素并构建Logistic回归模型,采用受试者工作特征(ROC)曲线分析Logistic回归模型对结直肠癌的诊断价值;进一步构建诊断结直肠癌的列线图并绘制校准曲线图进行内部验证。结果 癌症组患者血FIB、CEA水平及FOBT阳性率高于非癌症组,A/G、HB水平低于非癌症组,差异均有统计学意义(P<0.05)。将FIB、HB、FOBT、CEA进行多因素Logistic回归分析,建立Logistic回归模型作为4项指标联合检测模型:P=1/[1+e^(-(13.792-0.144×X_(1)+1.057×X_(2)+0.560×X_(3)+3.029×X_(4)))],其中X_(1)为HB水平,X_(2)为FIB水平,X_(3)为CEA水平,X_(4)为FOBT检测结果(阳性=1,阴性=0)。ROC曲线分析结果显示,当最大约登指数为0.841时,Logistic回归模型诊断结直肠癌的灵敏度为86.4%,特异度为97.7%,曲线下面积(AUC)为0.961(95%CI:0.933~0.988)。建立Logistic回归模型的列线图,并绘制校准曲线图,结果提示该模型的预测概率与实际概率具有较好的一致性。结论 基于FIB、HB、CEA和FOBT 4项指标构建的Logistic回归模型,对结直肠癌有着较好的诊断价值,也为临床对结直肠癌的早期筛查和诊断提供了一定的方向。Objective To evaluate the diagnostic value of combined detection of fibrinogen(FIB),hemoglobin(HB),carcinoembryonic antigen(CEA) and fecal occult blood test(FOBT) in patients with colorectal cancer,in order to provide reference for the diagnosis of colorectal cancer.Methods A total of 110 patients with colorectal cancer admitted and treated in the Colorectal Surgery Department of the General Hospital of Ningxia Medical University from January 2021 to June 2022 were selected as the cancer group,and 43 patients with benign colorectal disease during the same period were selected as the non-cancer group.The differences in the routine laboratory indexes(blood routine,biochemical routine,tumor markers,cytokines and FOBT) were retrospectively analyzed and compared between the cancer group and the non-cancer group.The Logistic regression was used to analyze the influencing factors of colorectal cancer and a Logistic regression model was constructed.The receiver operating characteristic(ROC) curve was used to analyze the diagnostic value of the Logistic regression model on colorectal cancer;a nomogram for diagnosing the colorectal cancer correlation was further constructed and a calibration curve chart was drawn for conducting the internal validation.Results The serum FIB and CEA levels and FOBT positive rate in the cancer group were higher than those in the non-cancer group,the A/G and HB levels were lower than those in the non-cancer group,and the differences were statistically significant(P<0.05).The multi variate Logistic regression analysis on FIB,HB,FOBT and CEA was conducted and the Logistic regression model served as the 4-item joint detection model:P=1/[1+e~(-(13.792-0.144×X_1+1.057×X_2+0.560×X_3+3.029×X_4))],where X_1,X_(2) and X_(3) were the HB,FIB and CEA levels respectively,and X_(4) was the FOBT detection result(positive=1,negative=0).The ROC curve analysis showed that when the maximum Jordan index of the Logistic regression model for diagnosing colorectal cancer was 0.841,the sensitivity of the model wa

关 键 词：结直肠癌 诊断价值 纤维蛋白原 癌胚抗原 粪便隐血试验 

分 类 号：R446.1[医药卫生—诊断学] R735.3[医药卫生—临床医学]