RESUMO
BACKGROUND AND AIMS: Surveillance programs are strongly recommended in patients with liver cirrhosis for early detection of HCC development. Six-monthly ultrasound sonography is the most reliable and commonly used technique, especially when associated with serum determination of α-fetoprotein, but different score systems have been proposed to overcome the unsatisfactory diagnostic accuracy of α-fetoprotein. The aim of this 12-year prospective study is to compare the gender, age, AFP-L3, AFP, des-gamma-carboxy prothrombin (GALAD) versus age, gender, bilirubin, albumin, and platelets and albumin-bilirubin scores in predicting HCC onset. APPROACH AND RESULTS: A cohort of 545 consecutive patients with compensated advanced chronic liver disease without suspected focal lesions was followed up every 6 months by liver imaging and α-fetoprotein to detect HCC occurrence. Harrell's C-index for censored data was employed to evaluate the performance of any parameters or scores helping to predict HCC development. ROC curve analysis showed that the GALAD score was more accurate in evaluating HCC development than albumin-bilirubin and age, gender, bilirubin, albumin, and platelets. The AUC ranged from 0.7268 to 0.6851 at 5 and 10 years, both in the total cohort and in the sub-cohorts (viral hepatitis, NASH, and alcohol). The HCC Risk model was constructed using univariate and multivariate Cox proportional hazard regression analysis, showing a strong association of GALAD with HR > 1, p < 0.05, in the total and sub-cohorts, and a better risk prediction in the alcohol cohort, both alone and standardized with other blood parameters. CONCLUSIONS: GALAD is the most reliable and accurate score system to detect HCC risk of development in patients with compensated advanced chronic liver disease.