Index of biochemical markers could reduce need for biopsy by half in HIV/HCV co-infected patients

Simon Collins, HIV i-Base

Liver biopsy is widely accepted as the gold standard for assessing hepatitis C (HCV) related fibrosis, but the procedure is invasive and can lead to complications, including haemorrhage and death. A paper published in a recent issue of AIDS suggests that an index of non-invasive tests may reduce the need for biopsy in over half the current cases.

Yves Benhamou and colleagues from Hopital Pitié-Salpetriere, Paris, looked at the predictive value of fibrosis from an index of non-invasive biochemical markers and whether this correlated with biopsy results in patients with HIV/HCV co-infection.

The cross-sectional, cohort study assessed 130 HIV/HCV-co-infected patients with a liver biopsy and serum samples for markers of liver fibrosis. The index incorporated age, sex, [alpha]2-macroglobulin, apolipoprotein A1, haptoglobin, bilirubin, and [gamma]-glutamyl-transpeptidase (GGT), derived using multivariate logistic regression, was compared with liver histology. HIV-specific indices including the CD4 cell count and HIV-RNA load were also constructed. The diagnostic values of the indices were compared using receiver operating characteristic (ROC) curves. Main outcome was measured by septal fibrosis (F2-F4) by the METAVIR classification.

By multivariate analysis, the most informative markers were [alpha]2-macroglobulin, apolipoprotein A1, GGT, and sex. The area under the ROC curve of the five-marker index was 0.856 ± 0.035 which is not significantly different from the HIV-specific indices. On a scale from zero to 1.00, the five-marker index had a positive predictive value of 86% for scores greater than 0.60, and a negative predictive value of 93% for scores of 0.20 or less.

The authors report that the accuracy of this index mean that these thresholds could reduce the need for liver biopsy by 55% while maintaining an accuracy of 89%. They also suggest that as “staging by biopsy is limited by semiquantitive scaling and sampling error, particularly in early stages and in cirrhotic patients with regenerative nodules”, that the estimates of the biochemical index could “actually provide a more accurate view of fibrogenic events occurring in the entire liver”.

Reference:

Yves Benhamou and Thierry Poynard et al. AIDS 2003; 17(5):721-725.

Comment

This study (and previous studies in mono-infected patients by the same group) is a landmark in the quest for non-invasive markers of liver damage in patients with chronic hepatitis C. Using an index based on sex, age, and five commonly available biochemical markers, the authors were able to predict a fibrosis score of F2 (a level at which treatment with interferon and ribavirin is generally recommended), with a fairly high accuracy. This index has been named ‘Fibrotest’. The authors conclude that using this index a liver biopsy could be avoided in 55% of the patients.

In an editorial that follows in the same issue of AIDS, Soriano et al argue whether liver biopsies, or assessment of the stage of fibrosis, are necessary at all in patients with HIV/HCV co-infection when treatment decisions are being made. They point out that the vast majority of patients with HIV/HCV co-infection will have appreciable fibrosis and therefore should be offered treatment. Furthermore, with the advent of combination therapy with Pegylated Interferon-alpha and ribavirin, where excellent results have been demonstrated in mono-infected patients, the major considerations in the decision making process should be CD4 counts and possible drug-interactions.

To put all this into perspective, let us consider the facts. In Western Europe and the USA, genotype 1 HCV accounts for almost two-thirds of all infections. HIV co-infection does accelerate the progression of HCV related fibrosis. With the advent of HAART, HCV related liver disease is now a leading cause of morbidity in co-infected patients. Treatment with interferon and ribavirin is curative. Liver biopsies (the gold standard for assessing liver damage) are uncomfortable and associated with risks in a minority of patients. Pegylated interferons in combination with ribavirin have improved the HCV clearance rates to over 75% in patients with genotypes 2/3 mono-infections and up to 45% in genotype 1 infection. Furthermore, responses, in terms of HCV viral load at week 12 of treatment, accurately predict long-term clearance rates. Having said all that, the early results of pegylated interferon-alpha and ribavirin in co-infected patients are largely disappointing, especially for patients with genotype 1 infections. Although final results from some of the larger randomised-controlled trials are still awaited, it is unlikely that we will see final HCV-clearance rates anywhere near the responses seen in mono-infected patients. Furthermore, there are significant drug-interactions with ddi and ribavirin. It is, however, still likely that the 12-week HCV viral load response will predict the final response.

So how does all this influence clinical practice? At the Royal Free Hospital in London, all patients with chronic genotype 2/3 HCV/HIV co-infection are offered therapy with pegylated interferon-alpha and ribavirin, bearing in mind drug-interactions and contra-indications to interferon-alpha and ribavirin. Therapy is stopped after 12 weeks if there has not been at least a two-log reduction in HCV viral load. Patients with genotype 1 HCV co-infections are less likely to respond to current therapy and therefore are offered a liver biopsy for accurate assessment of fibrosis. Those with moderate to advanced fibrosis are encouraged to try at least 12 weeks of therapy.

Once the biochemical markers in the Fibrotest index are routinely available, this will reduce the number of repeated liver biopsies (patients not on treatment and not having cirrhosis need repeated assessments every three to five years), although we would encourage at least one biopsy to ensure co-relation with the Fibrotest index, and encourage a biopsy for patients that fall in the ‘grey’ zone of scores between 0.2 and 0.6.