Histopathological evaluation of liver fibrosis: quantitative image analysis vs semi-quantitative scores. Comparison with serum markers

J Hepatol. 1998 Mar;28(3):439-46. doi: 10.1016/s0168-8278(98)80318-8.

Abstract

Background/aims: Liver fibrosis is mainly evaluated by qualitative histological examination. Although histological semi-quantitative scores and quantitative determination with image analysis are now possible, these methods have not been fully validated and compared. Therefore, we evaluated these two methods prospectively in 243 patients with chronic liver disease.

Methods: The semi-quantitative fibrosis score was evaluated by two independent pathologists, using the Knodell fibrosis score and a 6-grade score derived from the Metavir score; the area of fibrosis was measured by image analysis. The serum levels of hyaluronate, N-terminal peptide of procollagen III, laminin, transforming growth factor-beta1, alpha2-macroglobulin, apolipoprotein A1, PGA score and prothrombin index were measured.

Results: There was a good correlation between the semi-quantitative fibrosis score and the area of fibrosis (r=0.84, p<10(-4)). Using multiple regression analysis, the semi-quantitative score was predicted by the 8 serum markers with R2=0.69 (R2=0.59 for hyaluronate at the 1st step) while the area of fibrosis was predicted with R2=0.79 (R2=0.76 for hyaluronate at the 1st step), and the Knodell fibrosis score was predicted with R2=0.65 (R2=0.31 for hyaluronate at the 1st step).

Conclusions: The area of fibrosis, as determined by image analysis, and the semi-quantitative score are well correlated. However, for serum markers the correlation is higher with the area of fibrosis than with the semi-quantitative score. Other characteristics such as reproducibility, rapidity, simplicity, adaptability, and exhaustiveness also favor image analysis.

MeSH terms

  • Biomarkers / blood
  • Biopsy / methods
  • Hepatitis, Viral, Human / blood
  • Hepatitis, Viral, Human / pathology
  • Humans
  • Image Processing, Computer-Assisted*
  • Liver Cirrhosis / pathology*
  • Liver Diseases, Alcoholic / blood
  • Liver Diseases, Alcoholic / pathology
  • Observer Variation
  • Regression Analysis

Substances

  • Biomarkers