Z Gastroenterol 2015; 53 - P96
DOI: 10.1055/s-0035-1551784

A new algorithm for predicting the hepatocellular carcinoma occurrence in cirrhotic patients

S Bota 1, T Purevsambuu 1, F Hucke 1, T Reiberger 1, A Ferlitsch 1, H Hofer 1, W Sieghart 1, M Peck-Radosavljevic 1
  • 1Vienna HCC Study Group, Vienna Hepatic Hemodynamic Lab Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria

Aim: to develop an algorithm for predicting HCC incidence over time in a cohort of mixed etiologies of cirrhosis.

Methods: We retrospectively collected data of cirrhotic patients evaluated in our center between 2004 – 2014. In the same session HVPG measurements, laboratory parameters and AFP were determined. Patients with HCC at first presentation and those with a follow-up shorter than 90 days were excluded. HCC diagnosis was established by imaging techniques/biopsy during followed-up.

Results: We identified 940 cirrhotic patients that were evaluated with HVPG and laboratory blood tests and those which fulfilled the inclusion criteria were divided in 2 groups: training (379 patients evaluated between 2004 – 2009) and validation cohort (301 patients: 2010 – 2014). The HCC incidence in the training and validation cohort was: 8.7% and 6.3%.

Univariate analysis in the training cohort showed that presence of clinically significant portal hypertension (p = 0.007), platelet count< 100000cel/mm3 (p < 0.0001), AFP> 10 ng/ml (p = 0.002), presence of varices (p = 0.03), and age> 50 years (p = 0.04) were associated with HCC occurrence. The following factors were not associated with HCC occurrence: gender, etiology of liver cirrhosis, presence of diabetes mellitus, HIV coinfection, BMI, Child-Pugh class, MELD score, presence/history of ascites, presence/history of hepatic encephalopathy, aminotransferases values, bilirubin, albumin.

In multivariate analysis, platelet count< 100000/mm3 (OR = 4.36), age> 50 years (OR = 2.46) and AFP> 10 ng/ml (OR = 2.36) reached statistical significance.

Based on the odds-ratio, the following points were assigned: platelet count< 100000/mm3-2 points, age> 50 years-1 point and AFP> 10 ng/ml-1 point.

The patients were classified in three groups: group 1: 0 – 1 points, group 2: 2 – 3 points and group 3: 4 points.

The HCC incidence/year in the three groups in the training and validation cohort was: 0.6%, 3.2% and 9.2% (p < 0.0001) and respectively 0.3%, 5.5% and 9.2% (p < 0.0001).

Conclusions: The proposed algorithm can identify the patients with very low, moderate and high risk of HCC, which could be used for more targeted screening approaches.