432. Statistical and Data Mining Analysis to Identify Clinical, Biochemical and Pathological Features of Liver Fibrosis versus Metavir Score in a Cohort of 69106 Chronic Hepatitis C Patients in Egypt
Session: Poster Abstract Session: Hepatitis C
Thursday, October 27, 2016
Room: Poster Hall
  • Poster 2016-09-26.png (361.6 kB)
  • Background: In Egypt, the national program for HCV treatment has provided a rich pool of data which can led to the conduction of large population-based studies. Data mining analysis explores data to discover hidden patterns and trends.

    Aim: To identify independent clinical, biochemical and pathological features of liver fibrosis in a cohort of chronic HCV patients. And to develop a novel model to predict the stage of liver fibrosis

    Patients and Methods: This retrospective multi-centered-based study included pre-treatment clinical, laboratory, and histopathological data of 69106 chronic HCV Egyptian patients registered by the Egyptian National Committee for Control of Viral Hepatitis. Metavir scoring system of fibrosis F0-F4 was used as a gold standard to assess accuracy for the stage of fibrosis. Attributes with statistical significance correlation to liver fibrosis F2-F4 have been used to build a decision tree for predicting the stage of liver fibrosis

    Results: This study provided the demographic, biochemical, and histological characteristics for a cohort of chronic HCV patients presumably genotype-4 (Table 1). Minimal liver fibrosis (F0-F1) was observed in 32419 patients (46.9%), F2 in 25073 patients (36.3%), and advanced fibrosis (F3-F4) in 11615 patients (16.8%). There was discordances between FIB-4 and liver biopsy to diagnose advanced fibrosis (F3-F4) at cutoff value of 1.45 and 3.25 was 42.5% and 86.1% respectively, and discordance of 67.3% for APRI score at cutoff value of >1.

    The decision tree model showed that age was selected as the variable of initial split (most decisive), with optimal cut-off value of 37 years, the second important splitting attribute was AFP level  with optimal cut-off value of 6.5. Platelet count, other attributes as AST, and ALT glucose, BMI, albumin, bilirubin, and INR have less decisive role for prediction of fibrosis (Figure1).

    Conclusion: To our knowledge this multi-centered registry study has the largest sample size (69106 HCV patients) with liver biopsy histopathological results. The combination of age, AFP, platelet, AST improved diagnostic accuracy for fibrosis detection, and minimized the observed discordances between these scores versus liver biopsy.

    The decision tree model for prediction of fibrosis could be provided to doctors so they can know the probability of stage of fibrosis by simply entering simple data.

    Abubakr Awad, PhD Student in Computer Science1, Mahasen Mabrouk, MD2, Wafaa Elakel, MD2, Wahed Doss, MD2, Tahany Awad, MD2 and Samar Kamal, MD2, (1)Computer Science, Faculty of Computer Science, Cairo University, Cairo, Egypt, (2)Endemic Medicine and Hepatology Department, Faculty of Medicine, Cairo University, Cairo, Egypt


    A. Awad, None

    M. Mabrouk, None

    W. Elakel, None

    W. Doss, None

    T. Awad, None

    S. Kamal, None

    See more of: Hepatitis C
    See more of: Poster Abstract Session

    Findings in the abstracts are embargoed until 12:01 a.m. CDT, Wednesday Oct. 26th with the exception of research findings presented at the IDWeek press conferences.