P05 Combination of serum biomarkers for the diagnosis of hepatocellular carcinoma. (6th September 2011)

• Record Type: Journal Article
• Authors: Johnson, P; Teng, M; Pirrie, S; Palmer, D; Berhane, S; Stocken, D
• Is Part Of: Gut. Volume 60:(2011)Supplement 2
• Page Start: A3
• Page End: A3
• Publication Date: 2011-09-06
• Subjects: Gastroenterology -- Periodicals; 616.33
• DOI: 10.1136/gutjnl-2011-300857a.5
• Languages: English
• ISSNs: 0017-5749

Abstract

Abstract : Introduction: The most commonly used serum biomarker for the diagnosis of hepatocellular carcinoma (HCC) is α fetoprotein (AFP). However, AFP is also modestly raised in patients with chronic liver disease, particularly in those who are at a higher risk of developing HCC. A number of other circulating markers such as des-γ carboxy prothrombin (DCP) and the Lens culinaris agglutinin-reactive fraction of α-fetoprotein (L3) have been shown to have diagnostic discrimination which may, in certain situations, be superior to AFP. Aim: We therefore aimed to build a model that combined these biomarkers with baseline liver function tests, in order to identify those factors which predict HCC and might improve diagnostic efficacy of AFP alone. Method: A prospective study was specifically designed to assess the diagnostic efficacy of biomarkers. 461 serum samples were collected between 2007 and 2011: 252 were controls samples from patients with chronic liver disease with no evidence of HCC within 6 months of the sample; 59 were from patients with early HCC which was potentially curable and 150 were from patients with late HCC. AFP, AFP-L3 and DCP were measured in serum samples using a new micro total analysis system (Wako Chemicals GmbH, Neuss, Germany). Patients were classified into two groups (HCC, control) for the primary analysis. Independent predictors of HCC were identified from multivariable logistic regression analyses. The impact of AFP, DCP and L3 are reported based on the area under the receiver operator curve (AUC). Results: Univariate analysis showed AFP alone to have OR=2.96 (2.34 to 3.73 95% CI), AUC =0.8723, p<0.001. A multivariable model selected age, sex, AST, ALP and INR as independent predictors of HCC with an overall model AUC =0.8857. Including AFP, DCP and L3 to this model increased the AUC to 0.9489, 0.9290, 0.9254 respectively. The final model included age, sex, AFP, DCP and INR gave an AUC of 0.9661 (95% CI 0.95 to 0.98) for all patients and 0.944 (95% CI 0.914 to 0.974) for early HCC patients. Conclusion: A combination of serum markers may produce a clinically useful diagnostic test for HCC compared to AFP alone. Further validation of our model is underway on an independent dataset. These findings suggest that our model may enhance or replace the current screening approach for patients with liver cirrhosis, particularly since it appears to have efficacy even in patients with early, potentially curable, disease.