The United Kingdom (UK) population–based cohort study by Jackson et al.1 published in this issue of HEPATOLOGY provides new data on the mortality and incidence rates of malignancy in primary biliary cirrhosis (PBC). This long-term observational study, including nearly 1000 patients with a recorded diagnosis of PBC, compared the mortality in PBC patients and in a large matched general population cohort. The authors extracted and analyzed the information on PBC patients and control subjects from the UK General Practice Research Database, a longitudinal primary-care database set up in 1987 and covering a geographically dispersed sample of nearly 6% of the UK population.2 This approach allows one to minimize the risk of referral bias and is thought to better reflect the real world of clinical practice. Its disadvantage is the very limited number of clinical variables available for analysis in comparison with series collected from secondary or tertiary care centers with a special interest in the disease. In the present study, for example, no data about biochemical or histological features were analyzable, the severity of the disease being determined by the specific coding records of cirrhosis or portal hypertension. Likewise, the authors had to categorize patients as having regular or nonregular treatment according to prescription rates because the data were not complete enough to allow daily doses to be calculated. Despite this major drawback, the analysis of Jackson et al. provides interesting findings, especially regarding the influence of ursodeoxycholic acid (UDCA) therapy on mortality and the risk of primary liver cancer. Patients with PBC were found to have a 2.7-fold increased mortality rate in comparison with the UK general population. This finding is quite similar to that reported for the Northeast England cohort during the same follow-up period.3 However, in contrast, the mortality increase observed by Jackson et al. was lower in those patients regularly treated with UDCA versus untreated or short-term UDCA-treated patients (2.2-fold versus 2.7-fold), the groups being similar in terms of the disease severity at diagnosis. Although the difference between the mortality rates did not reach the level of significance, these data differ somewhat from those published in 2002 by Prince et al.,3 who found that UDCA was of no benefit and was even associated with a small increase in mortality. Furthermore, although no overall excess risk of malignancy was associated with PBC in Jackson et al.'s study, they showed that patients receiving the regular UDCA treatment had a lower increased risk of primary liver cancer than untreated patients (3-fold versus 8-fold). Although this is based on a small number of events, and even though there are overlapping confidence intervals, this further observation lends support to a preventive effect of UDCA therapy against cirrhosis development in patients with early-stage disease. Taken together, these results provide further observational data for a beneficial effect of UDCA on the long-term outcome in PBC.
Although UDCA is widely accepted as the standard medical treatment for PBC and currently is the only drug approved by the Food and Drug Administration in this disease setting,4 the evidence-based demonstration of its beneficial effect remains challenged by meta-analyses of short-term trials aimed at assessing UDCA efficacy in PBC. Most published randomized controlled trials (RCTs) have shown beneficial effects of UDCA on biochemical parameters. None of these trials, however, were of sufficient power to be able to find a significant decrease in the incidence rates of death or liver transplantation (LT). PBC is indeed a rare and usually slowly progressive disease. In this respect, the limited size and duration of the trials explain at least in part their inability to detect significant drug effects in terms of survival or survival without LT. A third parameter making trials even more complex to evaluate is the wide range of disease severity together with the fact that the magnitude of UDCA action appears to be inversely correlated to the disease stage. The issue can be summarized as follows: 2-4–year RCTs using survival as the endpoint are able to assess the progression of patients with cirrhosis or advanced disease in whom, unfortunately, UDCA has less chance to have a significant effect, but they are too short to be able to detect any beneficial effect in patients with earlier stages in whom UDCA probably exerts its maximal effect. The first demonstration of beneficial effects of UDCA on the survival of patients with PBC came from the combined analysis of 3 large RCTs.5 This study showed a significant effect of the drug in the subset of patients with moderate to severe disease that was compatible with a 32% reduction in the risk of death or LT. Since this publication, 5 meta-analyses, including 5-16 RCTs of UDCA, have been published.6–10 These meta-analyses came to different conclusions. Three of them (2 of which came from the same authors) concluded that there was an absence of a beneficial effect on the incidence of death, LT, and death or LT, despite a significant improvement in serum bilirubin, jaundice, and ascites.6, 7, 10 In contrast, the 2 others (1 of which was published in a correspondence form) showed that a long-term treatment with the optimal dose of UDCA could improve survival free of LT.8, 9 These conflicting results have been mainly attributed to the small sample size and the short duration (≤2 years) of several studies included in the former 3 meta-analyses and to the inclusion in these analyses of many trials administrating low doses of UDCA (≤10 mg/kg/day), which are currently known to be suboptimal.11 Finally, although none of the published RCTs were formally designed to assess specifically the effect of UDCA on histological stage progression, the combined analysis of 4 of these trials showed that 2 years of UDCA therapy delayed the progression of the histological stage in patients with early-stage disease, and this result supported the therapeutic efficacy of UDCA.12
Because the effect of UDCA on survival has not been proven by RCTs and meta-analyses beyond a doubt, further evidence derived from long-term–treated patient cohort studies has been expected. To date, several observational studies have been published.3, 13–18 Their conclusions are broadly consistent and support a long-term beneficial effect of UDCA. However, with the exception of the present publication, the UK experience appears to be to some extent different from that of other European countries. Indeed, although in the French, German, Spanish, and Greek observational studies a long-term treatment with UDCA was consistently found to improve survival free of LT in comparison with the predicted survival of untreated patients,13, 15, 17, 18 no such effect was observed in the UK series.3, 16 The observations derived from the non-British European cohorts are very consistent. All of them suggest that UDCA may normalize the survival rate of patients when it is given at early stages, that is, when patients have noncirrhotic, nonadvanced disease. Furthermore, the 10-year rates of survival without LT in these series were comparable, around 75%, whereas the survival rates estimated in the UK during the same period were about 50%. The estimated standardized mortality ratio in UDCA-treated patients was 1.8 in our French cohort,19 2.2 in the present study, and more than 2.9 in the series of Prince et al.3 What is so different on the other side of the Channel? First of all, the major observation that we can make is the surprisingly small reported proportion (37% in the study by Prince et al. and 43% in the present study) of British patients who received regular treatment with UDCA in the last 2 decades versus virtually all French patients during the same time (French Observatory of PBC, unpublished data, 2000). It appears that UDCA in the UK may have been preferentially prescribed to patients with more rapidly advancing disease, as suggested by Prince et al. Jackson et al.'s data support this hypothesis. The absence of a therapeutic effect in UK patients may therefore result from a selection bias, the patients with more severe disease and a lesser chance of response to UDCA being more likely to be treated. A second explanation may come from the study of the Northeast England cohort, in which the median daily dose of UDCA received between 1987 and 2000 was 7.5 mg/kg, that is, half of the therapeutic dose currently recommended to lead to biochemical improvement.11 Unfortunately, the data from the General Practice Research Database were not complete enough to confirm this observation. Nonetheless, taken together, these observations suggest that physicians in the UK have not been sufficiently convinced, motivated, or perhaps informed to widely and correctly prescribe UDCA to their patients in comparison with their European counterparts.
Finally, additional indirect support for a long-term beneficial effect of UDCA in PBC might come from retrospective studies of transplantation trends among patients with PBC. In one large series published in 2001, Liermann Garcia et al.20 showed that both the proportion and the absolute number of patients grafted for PBC were falling when both the prevalence and incidence of the disease were apparently rising.21 The data collected from the United Network for Organ Sharing in the United States also indicated that the number of LTs performed nationwide for patients with PBC decreased by an average of 7.2 cases per year from 1995-2004, despite a parallel increase in total LTs and no change in transplant rates for primary sclerosing cholangitis (PSC).22 The analysis of the French LT registry from the Agence de la Biomédecine23 strongly supports these observations. Figure 1 shows separately the trends of the French transplant activities for PBC and PSC for the period between 1990 and 2006. In agreement with the results of Lee et al.,22 the French data indicate that the absolute number of LTs for PBC has consistently being decreasing (by an average of 1.2 cases per year) for 15 years (P = 0.001), whereas the number of LTs for PSC has been increasing in parallel to the overall transplant activity. Similar trends have been observed for the number of PBC and PSC patients added to the French transplant waiting list (data not shown). Because UDCA therapy has not been shown to improve survival in patients with PSC,24, 25 the differences observed between the PBC and PSC transplant activities may constitute further evidence of the beneficial effect of UDCA on the natural history of PBC.