Advanced liver fibrosis may be predicted with more than 90% accuracy without the need for invasive biopsy by using a combination of laboratory markers, Spanish researchers reported last week at the Forty-Sixth Interscience Conference on Antimicrobial Agents and Chemotherapy in San Francisco.
The presence of liver fibrosis is an indicator that hepatitis C infection has begun to cause liver damage, and that treatment of hepatitis C infection is necessary in order to prevent further damage. Liver fibrosis can only be diagnosed by liver biopsy.
However, liver biopsy is a painful procedure that is understandably unpopular with patients, and takes up clinical time that might be devoted to other activities if a reliable non-invasive replacement could be found.
The search for non-invasive methods of predicting liver fibrosis has led researchers to test a number of combinations of laboratory markers, but so far none has definitively replaced liver biopsy.
Spanish researchers led by Dr Juan Berenguer of the Hospital Gregorio Marañón in Madrid took 296 consecutive patients with hepatitis C and HIV coinfection who were naive to hepatitis C treatment, and who had undergone a liver biopsy. Examining their laboratory data at the time of biopsy, the group identified clinical and laboratory variables that were associated with significant liver fibrosis. Patients were randomly assigned in a 70:30 ratio to an estimation and a validation group that were similar in all characteristics.
After identifying independent variables for significant liver fibrosis (F2 to F4, model HGM1) and advanced fibrosis (F3 and F4), the variables were combined in two equations, and compared with three recently reported non-invasive models for predicting liver fibrosis.
The HGM-1 index identified platelet count, glucose level and AST as independent predictors.
The HGM-2 index identified platelet count, INR, AST and alkaline phosphatase as independent predictors.
The indices were calculated by the equations below.
Using two cut-off values for each index, the investigators found:
- HGM-1 was able to predict the presence of F2-F4 fibrosis with 91% certainty at the higher cut-off (>0.848), but was less accurate at predicting the absence of significant fibrosis below a defined cut-off (<0.316) (68% certainty).
- HGM-2 was able to predict the presence of F3-F4 fibrosis with 95% certainty at the higher cut-off (>0.598) and the absence of F3-F4 fibrosis below the cut-off of 0.138 with 96% certainty.
- HGM-1 and 2 performed favourably in comparison to other algorithms, particularly in the case of HGM-2.
Berenguer J et al. Identification of liver fibrosis in HIV/HCV coinfected patients with a simple predictive model based on routine laboratory data. Forty-Sixth Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, abstract H-1885, 2006.