Statistics from Altmetric.com
We read with great interest the recent publication by Kim et al 1 showing that patients with untreated immune tolerant (IT) chronic hepatitis B with normal alanine aminotransferase (ALT) levels had significantly higher risks of hepatocellular carcinoma (HCC) and death/transplantation than treated immune active (IA) patients who had elevated ALT levels. These results indicate that ALT is not a sensitive surrogate marker for liver cell damage, and the IT phase of HBV infection cannot be considered fully benign.2 Chu and Liaw3 have challenged this view because they speculated that the IT group in that study might include IA patients who were in remission state after experiencing unrecognised minimal ALT elevations. Apparently, the highly dynamic nature of ALT perturbation makes it inadequate to act as a predictor for HCC.
We have developed and validated an easy-to-use scoring system for primary liver cancer (PLC) risk prediction. The development data set was set up in 1996 based on a community-based prospective cohort (qidong hepatitis B infection cohort (QBC)) established in Qidong, China.4 It contained 628 HBV surface antigen (HBsAg)-positive and 760 HBsAg-negative individuals. Participants with liver cirrhosis were excluded. After a median follow-up of 21 years, a total of 110 PLC cases occurred by the end of May 2017. Five potential predictors including age, gender, ALT, HBV e antigen (HBeAg) and HBV DNA were entered into a Cox regression model. Although in univariate analysis the baseline ALT level was significantly associated with PLC (p=0.0125), it failed to show significant HR in multivariate analysis (p=0.1157). This observation was consistent with the previous reports.5–7 Age, gender, HBeAg and HBV DNA levels were independent risk factors for PLC development (table 1). Based on these four variables, a 12-point risk score ‘AGED’ was generated. Point assignment for each predictor is shown in table 1. The model performances as indicated by the area under receiver operating characteristic curve (AUC) at 5, 10, 15 and 20 years were 0.76, 0.76, 0.79 and 0.80, respectively. The Χ2 of the Hosmer-Lemeshow tests was 6.15 (p=0.52).
The validation data set was initiated in 2007 with 1663 HBsAg-positive participants. Eighty-seven incident PLC cases were developed after a median follow-up of 10 years. AUCs at 5 and 10 years in the validation data set were 0.73 and 0.74, respectively. The χ2 of the Hosmer-Lemeshow tests was 7.99 (p=0.16).
The participants were classified into three risk groups based on an approximate tripartition of the AGED score. The low-risk group had 0–4 points, intermediate-risk group 5–9 points and high-risk group 10–12 points. The cumulative incidence probabilities of PLC were analysed in the low-risk, intermediate-risk and high-risk groups, and in the HBsAg-negative participants from QBC for a better comparison between the low-risk group and the HBsAg-negative population (figure 1). Compared with the low-risk group, the HRs in the high-risk group and intermediate-risk group were 20.3 (95% CI 9.69 to 42.53) and 4.67 (95% CI 2.33 to 9.36), respectively (both p<0.01), for development data set. Further stratification of the low-risk group showed that the group with 0–3 points had the similar PLC cumulative probabilities as compared with the HBsAg-negative group. In the validation data set, the HRs in the high-risk group and intermediate-risk group were 11.22 (95% CI 6.17 to 20.4) and 2.76 (95% CI 1.65 to 4.6), respectively (both p<0.01), compared with the low-risk group.
The AGED score provides a practical tool for PLC surveillance among HBV-infected individuals without liver cirrhosis. The main difference between our AGED score and the well-established Risk Estimation for Hepatocellular Carcinoma in Chronic Hepatitis B score8 is whether ALT is incorporated. The accuracy and applicability of AGED deserve further external validation in other HBV-infected populations who are at high risk of PLC.9
We would like to express our sincere gratitude to all of the cohort members whose long-term commitment in participating in the two cohorts made this prediction model possible.
Contributors CF screened the raw data, performed the statistical analysis and drafted the preliminary manuscript. ML and YG did HBV DNA tests and screened the raw data. TC, YS and JW maintained the derivation cohort. JC and JianhL maintained the validation cohort. YJ was responsible for frozen plasma samples. JianqL and GQ assisted in the acquisition of data. HT, JC and JG were responsible for the study concept and design, the acquisition, analysis and interpretation of data, critical revision of the manuscript for important intellectual content, and obtaining funding.
Funding This study was supported by the National Key Projects Specialized in Infectious Diseases (2017ZX10201201-008-003, 2017ZX10201201-006-002, 2012ZX10002-008-002 and 2012ZX10002-008-003), National Key Research Projects for Precision Medicine (2017YFC0908103), National Natural Science Foundation of China (81572312) and the Research Funding for Young Medical Talents of Nantong Municipal Commission of Health and Family Planning (WQZ2015007).
Competing interests None declared.
Patient consent Obtained.
Ethics approval The ethical committee of Qidong People’s Hospital/Qidong Liver Cancer Institute approved the study after a plenary session.
Provenance and peer review Not commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.