Serum fibrosis markers correlate with liver fibrosis stage in patients with advanced chr...(morphometry. On univariate analysis ...),Health

On univariate analysis, nearly all of the tested variables were independent predictors of cirrhosis, the authors report. They then conducted multivariate analyses and created a model that included TIMP-1, log HA and platelet count. The 3-variable model was significantly better than any of the individual serum fibrosis markers alone in estimating the presence of cirrhosis, they write. The model had an area under the receiver operating curve of 0.81 and was better at predicting cirrhosis than other published models.

The model would have correctly categorized 153 HALT-C patients as having a low likelihood of cirrhosis with 86% accuracy. An additional 146 subjects would have been categorized as having a high likelihood of cirrhosis with 73 % accuracy.

The serum fibrosis markers also correlated with the collagen content of biopsy samples, however not as closely as they did with the Ishak fibrosis scores. This suggests that the serum fibrosis markers more closely reflect the pattern of fibrosis determined by standard light microscopy than the quantity of hepatic collagen determined by computerized morphometry.

The study was limited by the unique nature of the HALT-C patient population and by the fact that the models were not tested in an external validation cohort. Still, the researchers conclude, on multivariate analysis, a 3-variable model consisting of TIMP-1, HA, and platelet count distinguished patients with non-cirrhotic CHC from those with cirrhosis. Also, this new model performed significantly better than other models based on routine laboratory tests, suggesting that serum fibrosis markers provide useful, incremental information in estimating disease stage in CHC.

'/>"/>