Author information

¹Biostatistics Program, Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA.

²Division of Gastroenterology and Hepatology, University of Michigan, Ann Arbor, MI.

³Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine and Science, Rochester, MN.

⁴Recanati/Miller Transplant Institute, Mt Sinai Medical Center, New York, NY.

⁵Division of Gastroenterology and Hepatology and Department of Epidemiology and Population Health, Stanford University, Palo Alto, CA.

⁶Division of Gastroenterology, Saint Louis University, St. Louis, MO.

⁷Department of Data Science, Dana Farber Cancer Institute, Boston, MA.

⁸Cancer Biomarkers Research Group, Division of Cancer Prevention, NCI, Bethesda, MD.

⁹Division of Digestive and Liver Disease, UT Southwestern, Dallas, TX.

¹⁰Biostatistics Program, Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA. Electronic address: jorge.marrero@pennmedicine.upenn.edu.

Abstract

Background & aims: Better surveillance tests for hepatocellular carcinoma (HCC) are needed. The GALAD score [Gender, Age, AFP-L3, AFP, and Des-carboxy-prothrombin] has been shown to have excellent sensitivity and specificity for HCC in phase two studies. We performed a phase three biomarker validation study to compare GALAD with AFP in detecting HCC.

Methods: This is a prospective study of patients with cirrhosis enrolled at seven centers. Surveillance for HCC was performed every 6 months at each site, and HCC diagnosis was confirmed per AASLD guidelines. Blood for biomarker research was obtained at each follow-up visit and stored in a biorepository. Measurements of AFP, AFP-L3, and DCP (des-gamma carboxyprothrombin) were performed in a FujiFilm laboratory by staff blinded to clinical data. The performance of GALAD in detecting HCC was retrospectively evaluated within 12 months prior to the clinical diagnosis. All analyses were conducted by an unblinded statistician in the EDRN data management and coordinating center.

Results: A total of 1,558 patients with cirrhosis were enrolled and followed for a median of 2.2 years. A total of 109 patients developed HCC (76 very early or early stage) with an annual incident rate of 2.4%. The AUC for AFP and GALAD within 12 months prior to HCC 0.66 and 0.78 (p<0.001), respectively. Using cutoff for GALAD of -1.36, the specificity was 82% and sensitivity at 12 months prior to HCC diagnosis was 62%. For comparison, performance of AFP at 82% specificity showed 41% sensitivity at 12 months prior to HCC diagnosis (p=0.001).

Conclusion: GALAD score, compared to AFP, improves the detection of HCC within 12 months prior to the actual diagnosis.