Editor's Note: Primary liver cancer ranks as the fourth most common malignancy and the second leading cause of cancer-related deaths in China, posing a severe threat to people's lives and health. Early detection, diagnosis, and treatment are crucial for improving treatment outcomes. In 2019, Dr. Feng Shen and Dr. Tian Yang research team from Eastern Hepatobiliary Surgery Hospital, collaborating with 11 renowned tertiary hospitals in China, developed the ASAP (Alpha-Fetoprotein, Age, Sex, and PIVKA-II) liver cancer risk assessment model, making a significant contribution to Chinese clinical practice.

Recently, the Feng Shen and Tian Yang research team from the Third Affiliated Hospital of Naval Medical University, together with the Guoyue Lv team from the First Hospital of Jilin University, published a multicenter case-control study comparing the diagnostic performance of the ASAP model and the GALAD model in detecting hepatocellular carcinoma (HCC) in chronic liver disease patients with different etiologies. The results showed that although the ASAP model excludes one laboratory variable (AFP-L3) compared to the GALAD model, it exhibits higher diagnostic performance for HCC associated with various chronic liver diseases. Hepatology Digest hereby reports the details as follows.

Research Background

HCC is the third leading cause of cancer-related deaths globally, often occurring in the context of chronic liver diseases such as hepatitis B virus (HBV), hepatitis C virus (HCV), alcoholic liver disease (ALD), and nonalcoholic fatty liver disease (NAFLD). Despite advancements in HCC treatment in recent decades, the long-term overall survival rate remains unsatisfactory, with only around 50% of patients surviving for one year after a late-stage HCC diagnosis. Therefore, early diagnosis is the only way to improve prognosis.

Serum tumor markers, due to their non-invasive, objective, and reproducible nature, have potential applications in the screening, monitoring, early detection, and prognostic assessment of malignant tumors. Three traditional HCC biomarkers—alpha-fetoprotein (AFP), AFP-L3 (Lens culinaris agglutinin-reactive AFP), and protein induced by vitamin K absence or antagonist-II (PIVKA-II, also known as DCP or des-γ-carboxyprothrombin)—have been approved by national regulatory authorities for routine clinical use. Among them, AFP is the most widely used. However, using the conventional threshold of 20 ng/mL, AFP has low sensitivity and specificity for detecting HCC, and AFP levels can also increase in HBV or HCV infected individuals without HCC. AFP-L3 and PIVKA-II alone also face similar limitations with suboptimal diagnostic performance.

Combining biomarkers with HCC risk-related clinical features can enhance diagnostic performance. The GALAD model, incorporating age, gender, AFP, AFP-L3, and DCP, achieves an area under the receiver operating characteristic curve (AUC) of 0.79 to 0.98 for HCC detection in HCV or NAFLD patients. Given that most HCC patients in China are HBV-related, the Tian Yang/Feng Shen research team established the ASAP model specifically for detecting HCC in HBV patients. Similar to the GALAD model, the ASAP model uses age, gender, AFP, and PIVKA-II but excludes AFP-L3 due to its low diagnostic performance in Asian patients. Data suggest that the ASAP model outperforms the GALAD model in diagnosing HBV-HCC (P=0.006). However, it remains unclear how the ASAP model performs in diagnosing HCC in patients with HCV, ALD, and NAFLD.

The research team from the Third Affiliated Hospital of Naval Medical University and the First Hospital of Jilin University conducted a multicenter case-control study utilizing a large cohort of chronic liver disease patients from 14 hospitals nationwide to compare the diagnostic performance of the ASAP and GALAD models in detecting HCC. Furthermore, subgroup analyses were performed in high-risk monitoring programs to compare the detection accuracy of both models, aiming to identify early HCC.

The research findings have been published online in Hepatobiliary & Pancreatic Diseases International, with co-first authors Dr. Liyang Sun and Prof. Nanya Wang from the First Hospital of Jilin University, and co-corresponding authors Prof. Feng Shen, Prof. Guoyue Lv, and Prof. Tian Yang.

Research results

This multi-center case-control study was completed in 14 hospitals across the country. Of the 1912 patients with chronic liver disease, with or without HCC, screened between May 2016 and December 2020, 970 patients, including 248 patients with HCC and 722 patients without HCC, were analyzed. Compared with the control group, the HCC group had a higher average age (58.6 years vs. 55.1 years, P<0.001), a higher proportion of males (79.0% vs. 72.3%, P=0.037), and a higher proportion of patients with cirrhosis (62.9% vs. 38.5%, P<0.001). In addition, the AFP, AFP-L3, PIVKA-II, ASAP score, and GALAD score in the HCC group were higher than those in the control group (P<0.001).

According to the different causes of chronic liver disease, patients were divided into four subgroups: HBV subgroup with 104 HBV-HCC patients and 367 HBV controls, HCV subgroup with 68 HCV-HCC patients and 132 HCV controls, ALD subgroup with 37 ALD-HCC patients and 76 ALD controls, and NAFLD subgroup with 39 NAFLD-HCC patients and 147 NAFLD controls. In the HBV and HCV subgroups, there are differences in gender and age between patients with HCC and chronic liver disease. In all subgroups, compared with the control subgroup of chronic liver disease, the proportion of patients with cirrhosis in the HCC subgroup was higher, and the median values of biomarkers and diagnostic model scores were higher (P<0.05).

Among the overall patients, the AUCs of AFP, AFP-L3, PIVKA-II, ASAP model, and GALAD model for detecting HCC were calculated. The results showed that the diagnostic performance of the ASAP model was the highest [AUC: 0.886, 95% confidence interval CI: 0.864-0.905], followed by the GALAD model with an AUC of 0.853, 95% CI: 0.829-0.875, P=0.001, PIVKA-II AUC: 0.819, 95% CI: 0.793-0.843, P<0.001, and AFP AUC: 0.750, 95% CI: 0.721-0.777, P<0.001. The diagnostic performance of AFP-L3 was the lowest with an AUC of 0.687, 95% CI: 0.657-0.716, P<0.001. The ASAP model had the highest sensitivity of 83% and specificity of 85%, compared to the GALAD model, which had a specificity of 89% and sensitivity of only 69%. In addition, when the specificity is fixed at 90%, the sensitivity of the ASAP model is the highest, at 73%, while the sensitivity of the GALAD model is only 67%.

According to the different causes of chronic liver disease, subgroup analysis was conducted. Among the four subgroups, compared with a single biomarker, the AUC of the ASAP model for detecting HCC was the highest. The AUCs in the HBV subgroup, HCV subgroup, ALD subgroup, and NAFLD subgroup were 0.88095%CI:0.8470.908, 0.89995%CI:0.8490.937, 0.89695%CI:0.824-0.945, and 0.87695%CI:0.820-0.920, respectively. The sensitivity of the ASAP model was the highest, reaching up to 85%, 87%, and 85% in the HBV subgroup, ALD subgroup, and NAFLD subgroup, respectively. The specificity was also high, reaching 86%, 84%, and 85%, respectively. In the HCV subgroup, although the GALAD model had the highest sensitivity of 84%, its specificity was only 73%. The ASAP model had a similar sensitivity of 78% and a higher specificity of 86%. When the specificity was fixed at 90%, the results in each subgroup were similar to those of the overall patient population.

According to different causes of chronic liver disease, 159 patients with early HCC were divided into four subgroups for analysis. Among all early HCC patients, the diagnostic performance of the ASAP model was the highest with an AUC of 0.863 and a 95% CI of 0.838-0.885, followed by the GALAD model with an AUC of 0.846 and a 95% CI of 0.821-0.870, although the difference between the two models did not reach statistical significance (P>0.05). When the specificity is fixed at 90%, the ASAP model shows a higher sensitivity of 70% vs. 65% compared to the GALAD model.

In all four subgroups, the AUC of the ASAP model for detecting early HCC was the highest, with AUCs of 0.85795% CI: 0.820-0.888, 0.87295% CI: 0.813-0.919, 0.84595% CI: 0.758-0.911, and 0.87195% CI: 0.813-0.915 in the HBV, HCV, ALD, and NAFLD subgroups, respectively. Compared with the GALAD model, the ASAP model had higher sensitivity in the HBV, ALD, and NAFLD subgroups, but there was no significant difference in the HCV subgroup. When the specificity was fixed at 90%, the ASAP model also had the highest sensitivity in all four subgroups.

Research Conclusion

This large-scale multi-center study, conducted in 14 hospitals across the country, is of great significance in evaluating and comparing the diagnostic performance of using AFP, AFP-L3, PIVKA-II, ASAP model, and GALAD model alone for detecting HCC in patients with chronic liver disease caused by four major causes: HBV, HCV, ALD, and NAFLD.

Among patients with chronic liver disease of different etiologies, the ASAP model, compared with the GALAD model, showed higher diagnostic performance AUC for identifying all stages or early HCC, despite using one less laboratory variable, AFP-L3, and had the highest sensitivity at a fixed specificity of 90%. These results suggest that the ASAP model has higher performance for screening and detecting early HCC in patients with chronic liver disease of different etiologies.

This study confirms that the ASAP model or GALAD model, which combines serum biomarkers and clinical characteristics, has higher diagnostic performance for detecting HCC in patients with chronic liver disease of different etiologies in China compared to single biomarkers such as AFP, AFP-L3, or PIVKA-II. The data of this study have important guiding significance and reference value for the screening, monitoring, diagnosis, and treatment of HCC in patients with chronic liver disease in China.

Therefore, the ASAP model shows higher diagnostic accuracy and cost-effectiveness, and is suitable for screening patients with chronic liver diseases caused by various etiologies, including HBV, HCV, ALD, and NAFLD, in high-risk populations of HCC in China, in order to detect HCC early.