Enhancement of AFP-Based Prediction of Onset of HCC by Using Other Laboratory Values

• Activity Number: 605
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Peter Richardson*+
• Companies: U. S. Veterans Health Admin
• Keywords: diagnostic medicine ; spline regression

Abstract

Serum alpha-fetoprotein (AFP) is a commonly used biomarker for detection of hepatocellular carcinoma (HCC). Its test characteristics are influenced by severity and activity of liver disease which are also reflected in serum levels of other lab tests. We investigated improving the ability of AFP to assess 6-month risk of HCC when information about alanine aminotransferase (ALT) values are incorporated in an AFP-based HCC detection algorithm. 11,721 patients with HCV+ cirrhosis with a total of 35,494 AFP tests in a national VA case registry were used in logistic modeling of the 6 month risk of HCC, based on continuous linear spline functions of log-transformed AFP and ALT values. AFP-bivariate and AFP-ALT multivariate models showed excellent calibration that was robust under cross-validation. 6-month HCC probabilities based on AFP alone within the range 20 to 120 ng/mL showed a 2-fold increase when ALT levels were restricted to 20 IU/L and a 5-fold increase at ALT = 10 U/L due to large AFP-ALT interaction at low ALT values which has been noted in clinical practice. We will also show the consequences of this for dichotomization-based clinical action at various HCC risk thresholds.