| Objective:To investigate the role of PIVKA-Ⅱ combined with AFP and AFP-L3 in patients with hepatocellular carcinoma in diagnosis of Hepatocellular Carcinoma(HCC),and to validate the application of GALAD model(consisting of Gender,Age,AFP-L3,AFP,and PIVKA-Ⅱ)and explore new models.Methods:A total of 641 cases were enrolled in the First Hospital of Jilin University from 2012 to 2015,including 254 with HCC(untreated and diagnosed for the first time),334 with chronic liver disease(CLD),and 53 healthy Controls(HC).The 334 CLD patients were divided into three subgroups: 188 with cirrhosis(Cirrhosis),96 with chronic hepatitis(CH)and 50 with other malignant liver tumor(OMT)which contained 23 with intrahepatic cholangiocarcinoma(ICC)and 27 with liver metastases(MT).Serum biomarkers AFP,AFP-L3 and PIVKA-Ⅱ were measured.We established a model called GALAD-C("C" means China)using the same variables(Gender,Age,AFP-L3,AFP and DCP)as GALAD and a new model called GAAP(consisting of Gender,Age,AFP,and PIVKA-Ⅱ)by logistic regression.Another 308 patients(including 169 HCC patients and 139 CLD patients)enrolled from 2015 to 2016 were used to validate the GAAP model.Results: Part one: comparison of median level of biomarkers in different groups1.The median level of AFP in HCC,Cirhhosis,Chronic Hepatitis,OMT and HC were 34.00 ng/m L,4.07 ng/m L,6.73 ng/mL,2.76 ng/mL and 2.74 ng/m L,respectively.HCC was significantly higher than all the other groups(all P values < 0.001);There was no significant difference between Cirrhosis and Chronic Hepatitis which both were higher than OMT and HC(all P values<0.001);The median level of PIVKA-Ⅱ in HCC,Cirhhosis,Chronic Hepatitis,OMT and HC were 201.4 mAU/mL,18.42 mAU/mL,23.84 mAU/m L,37.43 mAU/m L and 24.60 mAU/mL,respectively.HCC was significantly higher than all the other groups(all P values < 0.01);There was no significant difference between Cirrhosis,Chronic Hepatitis and HC.OMT was higher than Cirrhosis,Chronic Hepatitis and HC(all P values < 0.01).The positive rate of AFP-L3 in HCC was significantly higher than that in CLD and HC(37%,3% and 0,all P values <0.001).(AFP-L3 > 10% was defined as positive)2.As far as the median level of AFP was concerned,there was no significant difference between HBV related HCC and HCV related HCC(29.79 ng/mL vs.34.93 ng/mL),HBV related Cirrhosis and HCV related Cirrhosis(3.62 ng/m L vs.5.24 ng/m L);HCC(HBV related HCC,HCV related HCC)were higher than Cirrhosis(HBV related Cirrhosis and HCV related Cirrhosis)(all P values < 0.001);HBV related HCC had no difference with CHB(29.79 ng/mL vs.12 ng/mL);CHB was higher than that of CHC(4.2 ng/m L,P<0.05)and HBV related Cirrhosis(3.62 ng/mL,P<0.01).As far as the median level of PIVKA-Ⅱ was concerned: there was no significant difference between HBV related HCC and HCV related HCC(132.70 mAU/mL vs.435.5 mAU/mL),HBV related Cirrhosis and HCV related Cirrhosis(18.16 mAU/m L vs.20.02 mAU/m L),CHB and CHC(24.17 mAU/m L vs.22.66 mAU/mL);HCC was higher than that of Cirrhosis,CHB and CHC(all P values<0.001);there was also no significant difference between Cirrhosis,CHB and CHC.3.The median level of AFP,PIVKA-Ⅱ,GALAD-C model and GAAP model in the group of HCC with portal vein invasion were significantly higher than th ose of HCC without portal vein invasion(all P values<0.001).Part two: Correlation analysisThe correlation between AFP and ALT was identified(Spearman r= 0.2067,P<0.001),while there was no correlation between PIVKA-Ⅱ and ALT(Spear man r= 0.0229,P=0.591).There was a weak correlation between AFP and PI VKA-Ⅱ(Spearmanr were respectively 0.2869 and-0.0382,with P<0.001 and P=0.487 in HCC and CLD).AFP,PIVKA-Ⅱ,GALAD-C model and GAAP model were significantly correlated with the maximum diameter of the tumor(Spear manrwere respectively 0.1836,0.6381,0.5914 and 0.6002,all P values<0.01).Part three: the area under the receiver operating characteristic curve(AUC)1.AUC of GALAD-C model was significantly higher than that of AFP,PIVKAⅡ,AFP-L3,AFP+PIVKA-Ⅱ,AFP+AFP-L3 and AFP+ PIVKA-Ⅱ+AFP-L3(AUC value were 0.905,0.769,0.844,0.665,0.713,0.710,0.715,respectively,all P values<0.01)when HCC was differentiated from CLD,and GAAP model(AUC= 0.899)showed no difference between GALAD-C model in it(P=0.73).(“+” means “or”)2.AUC of GALAD-C model was significantly higher than that of AFP,PIVKAⅡ,AFP-L3(AUC value were 0.914,0.773,0.876,0.668,respectively,all P values<0.01)when HCC was differentiated from Cirrhosis,and GAAP model(AUC=0.906)showed no difference between GALAD-C model in it(P=0.68).3.AUC of GALAD-C model is significantly higher than that of AFP,PIVKA-Ⅱ and AFP-L3(AUC value were 0.817,0.702,0.765,0.560,all P values<0.01)when HCC within Milan criteria was differentiated from CLD,and GAAP model(AUC=0.817)showed no difference between GALAD-C model in it(P=1).4.AUC of GALAD-C model is significantly higher than that of AFP,PIVKA-Ⅱ and AFP-L3(AUC value were 0.985,0.919,0.884,0.680,all P values<0.01)when HCC were differentiated from HC,and GAAP model(AUC=0.984)showed no difference between GALAD-C model in it(P=0.91).Part four: validation testWhen HCC was differentiated from CLD using GAAP model with 308 cases in validation data,the sensitivity was 86%,specificity was 86%,False positive rate was 14%,False negative rate was 14%,correctly classified was 86%.ConclusionsAUC of GAAP model and the GALAD-C model showed no statistic significance and was better than the single serological markers.The validation data also confirmed that the GAAP model was of great value for HCC diagnosis in Chinese people. |
PDF Full Text Request