乙型肝炎相关性肝癌血清标志物诊断价值及支持向量机鉴别诊断模型的建立

The diagnostic value of serum biomarkers of hepatitis B-related HCC and the establishment of differential diagnosis model of SVM

目的:通过检测甲胎蛋白(AFP)、脱-γ-羧基凝血酶原(DCP)和高尔基体蛋白73(GP73)在乙型肝炎病毒(HBV)相关性肝细胞癌(HCC)患者血清中的表达水平,探讨单项及联合检测各项指标在HBV相关性HCC中的诊断价值。方法:建立支持向量机(SVM)鉴别诊断数学模型,用于不同阶段肝病的诊断与识别。选择就诊于医院的177例HBV患者,根据其病情将其分为HBV相关性HCC组(57例)、HBV相关性肝硬化组(52例)和慢性乙型肝炎组(68例);同期选择74名健康体检者纳入健康对照组,分别检测4组受试者的血清DCP、AFP及GP73水平,并对3项指标绘制受试者工作特征(ROC)曲线,获得ROC曲线下面积(AUC)及95%置信区间(95%CI)。结果:HBV相关性HCC组中血清DCP、AFP和GP73的表达水平显著高于健康对照组,且差异有统计学意义(U=304.5, U=345;P<0.05)和(t=-12.007,P<0.05);AUCDCP+GP73+AFP=0.997,95%CI(0.49~1.00);灵敏度、特异度、阳性预测值和阴性预测值分别为98.9%、98.6%、96.4%和98.6%。血清DCP、AFP及GP73水平3项指标联合检测效果好于单项检测,其差异有统计学意义(U=11.2,P<0.001)。HBV相关性HCC组与慢性乙型肝炎组比较AUC为0.7~0.9,与HBV相关性肝硬化组比较AUC<0.7。HBV相关性HCC组与HBV相关性肝硬化组SVM模型分类的特异度和灵敏度分别为90.91%和96.3%。结论:DCP、AFP及GP73指标联合检测可提高对HBV相关性HCC诊断的敏感性。SVM模型可用于不同阶段肝病的诊断与识别,可为HBV相关性HCC的诊断提供参考依据。

Objective: To explore the diagnostic value single and combined detection of each indicator in hepatocellular carcinoma (HCC) that was relative with hepatitis B virus (HBV) through detected expression level of serum alpha fetoprotein (AFP), DCP and GP73 of patients with HBV-related HCC. Methods: Through established mathematic model of differential diagnosis of support vector machine (SVM) to implement diagnosis and identification for hepatopathy at different stages. Total of 177 patients with HBV who received treatment in Beijing Tongren Hospital from May 2016 year to January 2018 year were collected and they divided into HBV-related HCC group (57 cases), HBV-related liver cirrhosis (LC) group (52 cases) and chronic HB group (68 cases) according to their conditions. Besides, 74 healthy persons who were confirmed by physical examination were included in healthy control group. The values of serum DCP ,AFP and GP73 of them were detected, respectively. And then, the receiver operator characteristic (ROC) curves of three indicators were drawn and area under curve (AUC) of ROC and 95% confidence interval (CI) were obtained. Results: The expression level of serum DCP, AFP and GP73 in HBV-related HCC group was significantly higher than that in health control group (U=304.5, U=345, t=-12.007, P<0.05). The results showed that AUCDCP+GP73+AFP=0.997 and 95% CI was (0.4930-1.0), and sensitivity, specificity, positive predictive value and negative predictive value were 98.9%, 98.6%, 96.4% and 98.6%, respectively. Combined diagnosis of three indicators was significantly better than single diagnosis(U=11.2, P<0.001). Compared with chronic HB group, the AUC of HBV-related HCC group was 0.7-0.9. And compared with HBV-related LC group, it was less than 0.7. Using SVM mathematical diagnosis model, the specificity and sensitivity of diagnosing classification of HBV-related HCC group and HBV-related LC group reached to 90.91% and 96.3%, respectively. Conclusion: The combined detection with serum DCP, AFP and GP73 can improve the diagnostic sensitivity of HBV-related HCC. And the SVM model can be used to diagnose and identify hepatopathy at different stages, and it can provide reference for diagnosis of HBV-related HCC.