In this issue (see page 121), Lueschner and colleagues extend previous reports regarding factors that predict biochemical normalisation in patients with primary biliary cirrhosis (PBC) being treated with ursodeoxycholic acid (UDCA).1 This study differs from others in having a much longer follow up period, an unusually large percentage of patients with early stage disease, and intensive evaluation during the follow up, including extensive biochemical, immunoserological, tumour marker, HLA typing, and histological evaluations.

They found, perhaps not surprisingly, that the less abnormal the tests initially, the more likely they were to become normal eventually on treatment and, perhaps even more obviously, the patients who had the best response after one year of treatment were most likely to have test values within the normal range when follow up was extended. They also noted histological improvement and the suggestion that symptoms improve in those patients in whom biochemical tests return to normal.

In Lueschner et al's study, there was a lack of correlation between the percentage of UDCA in serum bile acids and the biochemical response; this is different from previous reports.2 Furthermore, there was a lack of correlation between histological stage at entry and the biochemical response. This may be an artefact given that over 80% of the patients were at histological stages 1 and 2 at entry. Also unanticipated is the histological improvement reported in those patients in whom liver function tests returned to normal. This differs from the overall experience in other studies of the treatment of PBC with UDCA in which histological improvement was inconsistently described.3-6The slow rate of histological progression reported in this series, even in the incomplete responders, is also surprising. Based on modelling studies of untreated patients with PBC, substantially more than five of 47 patients would have been expected to develop histological progression over six years.7 One of the drawbacks of Lueschner et al's study, however, is the sparse description of how the histological grading was done. The methods section mentions a scale from 0 to 4 for several parameters, but the authors do not give details of how the score was derived. Furthermore, the histological data are mentioned in a single sentence and are not tabulated or otherwise presented, making it difficult to grasp fully the effects of treatment on the histological features of PBC.

Perhaps the most important unanswered question is the relation between normalisation of liver function tests and clinically relevant findings such as the development of cirrhosis and its complications. The meaning of this work could be as trivial as showing that those patients with the least abnormal liver tests before treatment have the least abnormal tests after treatment. It would be certainly important to know whether the biochemical normalisation correlated with clinically important outcomes. Given the early stages of disease in these patients, it is unlikely over even a five year period that these patients would die or require liver transplantation or develop the complications of liver disease which the authors have outlined so carefully. Perhaps another measure would be to assess the correlation between biochemical changes and changes in the Mayo Risk Score. This is mentioned only in passing, the authors stating in the discussion that the Mayo Risk Score or European Risk Score (presumably calculated before treatment began) did not predict the biochemical response to UDCA. Although there are hazards in using surrogate markers, two separate series have shown that the Mayo Risk Score predicts prognosis accurately in patients being treated with UDCA.8 9 The results of this calculation for the patients in Lueschner et al's study would be of interest.

Early measures of treatment response will be particularly important when other forms of therapy become available, particularly when they are to be used as adjuncts to treatment with or in place of treatment with UDCA. Lueschner et al's study suggests that patients with almost normal test results after one year were most likely to go on to complete biochemical normalisation, although as discussed earlier, the clinical implications of this are truly unclear. This study does help to confirm the value of the initial response of alkaline phosphatase activities to UDCA treatment as a potential determinant of candidacy for adjuvant therapy, as has been reported recently.9 In the previously reported study, higher alkaline phosphatase activities after at least six months of UDCA therapy correlated with a poorer clinical outcome. Future studies evaluating the issue of predicting response to UDCA will need to focus both on the biochemical response and on carefully described histological findings and other clinically relevant end points.