BEIJING, CHINA—Researchers from Peking University and collaborating hospitals across China have developed and validated GADA, a simplified model for detecting hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC). Derived from the established GALAD framework, GADA combines gender, age, des-gamma-carboxy prothrombin (DCP; also known as PIVKA-II), and alpha-fetoprotein (AFP). In more than 23,000 participants, it showed strong diagnostic performance for both overall and early-stage HCC. The study is forthcoming in the Chinese Medical Journal.
Chronic HBV infection causes more than 40% of HCC cases worldwide and is associated with approximately 86% of HCC cases in China. The epidemiology of HCC in China therefore differs substantially from that in many Western cohorts, where hepatitis C, alcohol-related liver disease, and metabolic liver disease contribute more prominently.
The GALAD model integrates gender, age, AFP, AFP-L3%, and DCP and has performed well in several Western populations. However, its performance required reassessment in HBV-associated chronic liver disease. AFP-L3% also adds measurement complexity, cost, and sample-volume requirements, and is not readily available in all clinical settings. The study, led by Professor Fengmin Lu was published online on August 17, 2026, in the Chinese Medical Journal .
The investigators analyzed 2,190 adults with HBV-related chronic liver disease at Mengchao Hepatobiliary Hospital of Fujian Medical University. Participants were randomly assigned to a modeling cohort (70%; n = 1,533) and an internal validation cohort (30%; n = 657). A separate prospective external validation cohort enrolled 20,908 participants at ten hospitals across China, including 905 patients with newly diagnosed HCC and 20,003 patients without HCC (17,802 with chronic hepatitis B and 2,201 with HBV-related cirrhosis). Altogether, 23,098 participants were included.
AFP and DCP provided complementary diagnostic information, whereas AFP-L3% contributed limited additional utility in this HBV-focused population. The researchers therefore developed the four-variable GADA score:
Z = –11.111 + 1.122 x gender + 0.073 x age + 2.660 x log10(DCP) + 0.578 x log10(AFP)
Gender is coded as 1 for male and 0 for female.
Cohort |
AUC (95% CI) |
Sensitivity |
Specificity |
Modeling |
0.940 (0.927–0.952) |
82.57% |
91.53% |
Internal validation |
0.935 (0.914–0.953) |
82.04% |
92.65% |
External validation |
0.919 (0.916–0.923) |
86.00% |
81.88% |
In the modeling cohort, GADA outperformed the original GALAD model (AUC 0.940 vs. 0.907). In the external validation cohort, it also outperformed AFP alone (AUC 0.748) and DCP alone (AUC 0.841). Center-specific AUCs ranged from 0.895 to 0.977, supporting robustness across geographically and clinically diverse settings.
For early-stage HCC meeting the Milan criteria, GADA achieved AUCs of 0.858 in the modeling cohort, 0.868 in the internal validation cohort, and 0.903 in the external validation cohort. In the external cohort, sensitivity was 88.16% and specificity was 74.50%.
GADA was designed for the setting in which the clinical need is greatest in China: distinguishing HBV-related HCC from chronic hepatitis B and HBV-related cirrhosis. By removing AFP-L3% while preserving strong diagnostic accuracy, the model may reduce testing complexity and improve accessibility, including in hospitals with more limited laboratory resources.
The model should be viewed as an adjunct to established HCC surveillance and diagnostic pathways, not as a replacement for imaging or clinical assessment. Prospective evaluation in routine surveillance populations and calibration of operational thresholds across laboratories will be important before broad implementation.
DOI: 10.1097/CM9.0000000000004244
Prof. Fengmin Lu is Director of the Peking University Infectious Disease Research Center, Professor in the Department of Microbiology at Peking University Health Science Center, and a jointly appointed Professor at the Center for Infectious Diseases, Peking University Third Hospital. His research focuses on the molecular mechanisms of HBV-related hepatocarcinogenesis, occult HBV infection, and biomarkers. He led the development and multicenter validation of the GADA model. He has led more than 20 national research projects, received two Second-Class National Science and Technology Progress Awards, published nearly 200 papers, and holds 25 authorized invention patents, seven of which have been commercialized.
This work was supported by the National Natural Science Foundation of China (No. 82272433) and the Fujian Provincial Key Laboratory of Hepatic Drug Research (No. 2022-YF-0050).