New methods to detect liver fibrosis severity and predict clinical outcomes in chronic hepatitis C infection

    Research output: ThesisDoctoral Thesis

    482 Downloads (Pure)

    Abstract

    Chronic hepatitis C (CHC) is a slowly progressive disease, with an early longterm asymptomatic phase. To accurately predict fibrosis severity and the risk ofadverse clinical outcomes is a major challenge for individual management ofCHC patients. While liver biopsy remains as the reference standard, a largenumber of non-invasive methods emerged as surrogate markers for liverfibrosis assessment and clinical outcome prediction. This thesis developed andvalidated the optimum method of image analysis to measure collagen proportional area (CPA) of liver biopsy. Novel serum models have also beendeveloped to predict CPA and also to directly predict liver related clinical outcomes.

    880 CHC patients who had a liver biopsy done in Sir Charles Gairdner Hospitalfrom 1992 to 2012 were included in the thesis. Detailed follow up informationwas obtained using Western Australia data linkage unit for each patient. Clinicaldata including serum test results, age and gender were obtained from hospitaldatabase. The optimum method of CPA measurement was determined. Usingthe optimised method, CPA stage [C1: 0%-5% (normal), C2: 5%-10% (minimal),C3: 10%-20% (moderate), C4: >20% (severe)] was able to stratify risk betterthan Metavir stage with a significant difference in HCC free survival betweeneach consecutive CPA stage. CPA stage remained as an independent predictorfor liver related death and HCC after adjusting for Metavir stage and age. CPAstage was also significantly associated with the risk of liver related death incirrhotic patients. Additionally, CPA measurement has the potential to use smallbiopsies that were insufficient for histopathological staging. A serum model (CPAscore) was developed to detect liver fibrosis severity using CPA as thereference standard. CPAscore included HA, α2-macroglobulin and plateletcount and it was closely correlated with actual CPA values with the model fit (Rsquare) of 0.46. CPAscore achieved satisfying accuracy to detect those patientswith CPA larger than 10% and 20% with AUROC of 0.82 and 0.94 respectively.Using cut points of 8.7 and 10.7, those patients with CPA larger than 10% and20% were detected respectively with both high sensitivity and specificity. Threeserum models [Liver Outcome Score (LOS)] were developed to directly predictliver related death, hepatocellular carcinoma (HCC) and liver decompensationrespectively. LOS panels showed a high accuracy to predict five year liverrelated death, decompensation and HCC with an AUROC of 0.95, 0.90 and0.95 respectively. Using the defined cut points, those patients categorised in thehigh risk group for liver related death, HCC development and decompensationhad an annual incidence rate of 12.6%, 6.27% and 4.54% respectively and wassignificantly higher than that of low or moderate risk group (p<0.001).

    In conclusion, this project it is the largest study that comprehensively evaluatedthe ability of CPA to predict clinical outcomes. Moreover, four serum modelswere developed to predict CPA value, the risk of liver related death, HCC andliver decompensation respectively. After validation, these serum models can beused as standard tests in clinical practice and guide individual patientmanagement of chronic hepatitis C infection.

    Original languageEnglish
    QualificationDoctor of Philosophy
    Publication statusUnpublished - 2015

    Fingerprint Dive into the research topics of 'New methods to detect liver fibrosis severity and predict clinical outcomes in chronic hepatitis C infection'. Together they form a unique fingerprint.

    Cite this