Diagnostic performance of the GALAD model and comparable algorithms in early-stage hepatitis B virus-related hepatocellular carcinoma

OUYANG Weiqing; YU Guangchao; SUI Hong; ZHENG Biying

doi:10.3969/j.issn.1009-0460.2025.04.007

Abstract：Objective To evaluate the diagnostic value of the hepatocellular carcinoma(HCC)diagnostic model GALAD and its derived models,GALAD-US and GALAD-C,in the early detection of hepatitis B virus-related HCC(HBV-HCC)in patients with chronic hepatitis B(CHB).Methods A total of 1110 CHB patients treated at Dongguan Kanghua Hospital from January 2019 to May 2022 were enrolled and divided into an HBV-HCC group(224 cases,confirmed by pathology or imaging)and a non-HCC group(886 cases).After excluding 33 patients in the HBV-HCC group who did not meet the Milan criteria and had undergone anticancer treatment,191 cases were included in the analysis,comprising an untreated HBV-HCC subgroup(123 cases)and an early-stage HBV-HCC subgroup meeting the Milan criteria(68 cases).The non-HCC group was further divided into a CHB-only subgroup(637 cases)and an HBV cirrhosis-only subgroup(249 cases).Receiver operating characteristic curve analysis was performed to compare the diagnostic performance of individual markers-alpha-fetoprotein(AFP),alpha-fetoprotein-L3 isoform ratio(AFP-L3%),des-gamma-carboxy prothrombin(DCP)—and their combination,as well as the GALAD,GALAD-C,and GALAD-US models in the overall HBV-HCC group and its subgroups.Results In the overall HBV-HCC versus non-HBV-HCC comparison,The GALAD-US model demonstrated the highest diagnostic efficacy with an area under the receiver operating characteristic curve(AUC)of 0.951,showing no statistically significant difference compared with the GALAD-C model(AUC=0.946,P>0.05).In the Milan criteria early-stage HBV-HCC versus non-HBV-HCC comparison,the GALAD-C model showed the highest diagnostic performance(AUC=0.944),significantly outperforming the GALAD model,individual AFP,AFP-L3%,DCP,and their combination(AUCs:0.905,0.758,0.702,0.849,and 0.884,respectively;P<0.05).Similarly,in the Milan criteria early-stage HBV-HCC versus HBV cirrhosis-only comparison,the GALAD-C model achieved the highest diagnostic efficacy(AUC=0.905),significantly surpassing the GALAD-US,GALAD,individual AFP,AFP-L3%,DCP,and their combination(AUCs:0.859,0.859,0.728,0.682,0.841,and 0.859,respectively;P<0.05).Conclusion The GALAD-C model is the most effective tool for diagnosing early-stage HBV-HCC,particularly in distinguishing between early HBV-HCC and HBV cirrhosis alone.