血清SPINK1/SPP1联合诊断原发性肝癌的效能评价

Evaluation of the performance of SPINK1/SPP1 in diagnosis of hepatocellular carcinoma,alone or in combination

目的探讨丝氨酸蛋白酶抑制剂因子Kazal 1型(SPINK1)、骨桥蛋白(OPN/SPP1,下称SPP1)单独或联合模型对于原发性肝癌的诊断效能。方法收集血清样本419例,其中正常对照(Normal)93例,慢性乙肝肝炎(chronic hepatitis B,CHB)72例、乙肝肝硬化(liver cirrhosis,LC)77例及原发性肝癌177例。通过ELISA方法检测SPINK1和SPP1水平。通过显著性检验判断SPINK/SPP1及主要实验室检验。在各组间是否存在统计学差异;通过Logistic回归分析建立包括SPINK/SPP1的联合诊断模型,通过灵敏度、特异性、整体准确率等指标评价模型的诊断效能。结果肿瘤组SPINK1及SPP1血清水平均明显高于其他各组(P<0.0001)。ROC曲线分析显示,甲胎蛋白(AFP)、SPINK1及SPP1的曲线下面积(AUC)分别为0.798(0.755~0.845)、0.951(0.931~0.971)和0.939(0.916~0.963),基于SPINK1/SPP1的联合诊断模型可达到0.972(0.957~0.986),AFP明显低于其他各组(P<0.0001),SPINK1和SPP1的AUC差异则无统计学意义(P>0.05)但均明显低于联合模型(P=0.0429和P<0.001)。四个指标AFP、SPINK1、SPP1及联合模型分别取截断值14.28、7.39、8.14及0.403时,灵敏度、特异性、准确率分别为0.678、0.864、0.785(AFP),0.949、0.851、0.892(SPINK1),0.915、0.855、0.877(SPP1)及0.938、0.918、0.926(联合模型)。多组间比较差异具有统计学意义(渐进P<0.001)。两两比较,AFP与其他三组间差异均有统计学意义(P<0.0001),而SPINK1(P>0.05)、SPP1(P>0.05)及联合模型(P>0.05)各组间均无统计学差异。结论血清SPINK1/SPP1水平单独或联合应用对于原发性肝癌的诊断效能明显优于血清AFP,有望成为原发性肝癌的新型肿瘤标志物。

Objective To evaluate the performance of SPINK1/SPP1 in diagnosis of hepatocellular carcinoma(HCC)alone or in combination.Methods A total of 419 serum samples were collected and divided into four groups:normal control(n=93),chronic hepatitis B(CHB)(n=72),HBC related liver cirrhosis(LC)(n=77),and hepatocellular carcinoma(HCC)(n=177).Serum concentrations of SPINK1 and SPP1 were determined by ELISA kits.All parameters were first analyzed by significance tests among the groups.To distinguish tumors from non-tumors,a combination model was generated by multivariable binary Logistic regression using a comprehensive control group which consisted of the normal,the CHB and the LC groups.The performance of each indicator was judged by comparison of AUC,sensitivity,specificity and accuracy.Results The serum levels of SPINK1 and SPP1 were both significantly higher in HCC group than in all the others(P<0.0001).ROC curve analysis showed that AFP,SPINK1 and SPP1 achieved AUC of 0.798(0.755-0.845),0.951(0.931-0.971)and 0.939(0.916-0.963),respectively.The combination model based on SPINK1 and SPP1 reached the highest AUC of 0.972(0.957-0.986).AUC of AFP was significantly lower than that of the other indicators(P<0.0001);meanwhile,AUC of SPINK1and SPP1,with no significant difference(P>0.05),was lower than that of the combination model(P=0.0429 and P<0.001,respectively).With the cutoff of 14.28,AFP established sensitivity,specificity and accuracy of0.678,0.864,and 0.785,which were significantly lower than those in the other groups,with those of 0.949,0.851and 0.892(SPINK1,cutoff=7.39);0.915,0.855 and 0.877(SPP1,cutoff=8.14);0.938,0.918 and 0.926(combination model,cutoff=0.403).AFP differed significantly from that in the other three groups(P<0.0001).For comparison of SPINK1 or SPP1 with combination model,the differences were not statistically significant(both P>0.05).Conclusion The data identified SPINK1 and SPP1 as novel tumor biomarkers with greater robust efficiency than the currently used AFP for detection of hepatocellular carcinoma,alone or in combination.