Statistics from Altmetric.com
We read with great interest the article by Ryder and the Trent Hepatitis C Study Group (Gut 2004;53:451–5) addressing the issue of fibrosis progression in untreated patients with chronic hepatitis C. After examining a large series of paired liver biopsies (median inter-biopsy interval 2.5 years), the authors concluded that even in patients with a “histologically mild” presentation, chronic hepatitis C is characterised by progressive fibrosis. Histological assessment of liver fibrosis in chronic liver diseases is an area of interest to us and we wish to raise a methodological issue concerning Ryder’s study.
While many attempts have been made to standardise non-invasive tests for predicting liver fibrosis, biopsy sampling is currently considered the standard reference for grading and staging chronic hepatitis. Biopsy sampling error and sample size have proved the major sources of bias in liver fibrosis assessment and both variables should be taken into account when the outcome of histology guides patient management decisions.
Two studies1,2 have recently demonstrated that the traditional concept of “adequate liver biopsy sample” (core biopsy 1.5 cm long and/or containing 4/5 portal tracts) is no longer acceptable. Both studies consistently demonstrated that the stage of disease is underrated in biopsy samples shorter than 2–2.5 cm and/or thinner than 1.4–1.2 mm. More recently, a well designed study by Brunetti and colleagues3 showed that fine needle biopsies also underestimate the stage of chronic hepatitis C, as revealed by large needle biopsies. The critical factor in staging fibrosis is the number of complete portal tracts included in the sample, the range of 11–15 complete portal tracts being the minimum requirement for a reliable histological assessment.1 In practice, this means that when paired biopsies are used to assess fibrosis progression, biopsy size and/or number of complete portal tracts has to be comparable in the two samples considered: any progression or regression of fibrosis may be biased due to the inconsistent size of the paired liver biopsy samples (that is, initial small versus larger follow up biopsy, or large initial versus smaller follow up biopsy).
In the Ryder study, the threshold for adequacy was the presence of more than five portal tracts and the authors did not say whether the paired biopsies were comparable in terms of the number of portal triads. Fibrosis regression (10%) was interpreted as “presumably” being due to “a combination of observer error and sampling variation”. We agree that this is a possible interpretation of “regressive” fibrosis but we would suggest that a similar interpretation should also be considered for “progressing” fibrosis.
Our concern is of some practical importance as Ryder’s study demonstrates that fibrosis progresses, in less than three years in 33% of untreated hepatitis C patients (including those with persistently normal alanine aminotransferase). This means that the strategy that excludes subjects with mild disease from treatment (as recommended by current guidelines) needs to be reconsidered.
The size of liver biopsy samples is feasibly one of the clinical variable pertaining to in vivo studies and it would be unreasonable to disregard liver biopsies that fail to fulfil current adequacy criteria entirely. None the less, the considerable bias that “inadequate” biopsy material can introduce makes it necessary to pay more attention to ensure an exhaustive description of the analysed samples (too often neglected in the “Methods section” of otherwise valuable articles). This methodological recommendation applies both to pathologists involved in clinical trials and to manuscript reviewers.4
Drs Guido and Rugge make important points about the interpretation of liver biopsy if the sample is small.
Our study Gut 200453451-5 was accepted for publication before the Colloredo article was published in the Journal of Hepatology and we would agree that the standards for accepting a liver biopsy as adequate have changed in the light of their data.
With respect to our study, we can confirm that the median number of portal tracts per biopsy was 15 in the second biopsy that showed regression with no biopsy having less than 11. Analysis of size and portal tract number in the groups of patients who progressed, remained stable, or regressed showed no difference. In particular, the second biopsy sample in regressors contained a higher median number of portal tracts than the index biopsy. Using the criteria established by Brunetti et al, only six biopsies from the whole cohort would have been characterised by them as inadequate, and omitting these patients from the analysis does not alter the results.
We therefore feel that the biopsy size in our study was adequate and the conclusions remain valid.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.