Clinical Scoring System for Prediction of Long-term Risk for Hepatocellular Carcinoma Among Hepatitis C Virus Infected Patients

Mei-Hsuan Lee1, Hwai-I Yang2,3, Sheng-Nan Lu4, Chin-Lan Jen1, San-Lin You1, Li-Yu Wang5, Yong Yuan6, Gilbert L’Italien6,7, and Chien-Jen Chen1,8 for the REVEAL-HCV Study Group

Affiliations
1Genomics Research Center, Academia Sinica, Taipei, Taiwan
2Molecular and Genomic Epidemiology Center, China Medical University Hospital, Taichung, Taiwan
3Graduate Institute of Clinical Medical Science, China Medical University, Taichung, Taiwan
4Department of Gastroenterology, Kaohsiung Chang-Gung Memorial Hospital, Kaohsiung, Taiwan
5MacKay College of Medicine, Taipei, Taiwan
6Global Health Economics and Outcomes Research, Bristol Meyer Squibb, Princeton, NJ, USA
7Yale University School of Medicine, New Haven, CT, USA
8Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan

Background and Aims: We aimed to develop a simple and non-invasive tool for prediction of long-term risk for hepatocellular carcinoma (HCC) among hepatitis C virus (HCV) infected patients.

Methods: The study included 975 anti-HCV seropositives who were seronegative for HBsAg. They were adults aged between 30-65 years old and were enrolled during 1991-1992. At study entry, the seromarkers including serum HCV RNA levels, alanine aminotransferase (ALT) and asparate aminotransferase (AST) levels, HCV genotypes were examined. The study subjects were followed for their newly developed HCC by computerized linkage with the National Cancer Registration System in Taiwan during 1991 to Dec., 31, 2008. There were three models developed: model 1 included age, ALT, AST/ALT ratio (AAR), and the presence of cirrhosis, model 2 included the risk factors in model 1 plus serum HCV RNA, and model 3 added HCV genotypes in additionally. The Cox’s proportional hazards models were utilized to estimate the regression coefficients of each risk factor for HCC. A coefficient could be converted to an integer as a simple risk score. The risk for 10- and 15-year risks for HCC based on the sum of risk score of each risk factor could be calculated. The area under receiver operating curve (AUROC) was used to evaluate the performance of each model.

Results: All of the risk factors we included in prediction models were significantly associated with HCC. The higher the total risk score, the higher the risk of developing HCC. For those with the highest score, the 10-year and 15-year HCC risk in full model was 82.9% and 98.7%, respectively. The AUROCs of each model for 10-year and 15-year risk prediction were 0.8.

Conclusions: The clinical scoring system provided information for finding patients who need more intensive care. However, an external cohort for validating the performance of accuracy is necessary.

（该文在第22届亚太肝脏研究学会年会上首次发表，该文第一作者台湾科学院李美璇博士获得第22届亚太肝脏研究学会年会年轻研究学者奖）