Various anti-inflammatory or immunosuppressive drugs have been tested to treat primary biliary cirrhosis (PBC). None has provided a satisfactory response, and none is routinely recommended. This is a predicament for a disease in which lymphocyte superactivation leads to destruction of liver parenchyma. Are the investigators liable because of methodological flaws in their trials, or is it their ignorance of the true course and nature of PBC? The answer to both questions is obviously affirmative. However, negative clinical trials have hidden pearls, which have provided us a wealth of information during the last two decades.

Investigators planning a clinical trial need to agree on a primary study question. Sample size, study duration, appropriate selection of homogeneous patients, i.e. the quality of a clinical trial, depends on this key question. The design and endpoint are necessarily based on an accurate picture of the natural history and of factors predictive of survival or progression of the disease. In the 1980s, it was generally assumed that early histological stage and asymptomatic PBC did not justify treatment. Even recently, it was debated whether asymptomatic patients have a progressive disease and a life expectancy shorter than a matched control population.1 Based on more recent studies, it is clear that PBC is a progressive ailment with an estimated median survival of 10 to 15 years.2 Its natural course can be broken down into three distinct phases: (1) an early phase characterized by ongoing inflammation and destruction of the interlobular bile ducts accompanied in more than half of the cases by lobular inflammation, mainly under the picture of periportal lymphocytic piecemeal necrosis; (2) a second phase marked by irreversible loss of bile ducts, remodeling parenchyma with nodular hyperplasia, extensive fibrosis, and cirrhosis; and (3) a third phase occurring when serum bilirubin reaches 100 μmol/L and coined the terminal phase as the mean survival of such patients is less than 3 years.3

Several ancillary studies of randomized trials allowed quantification of the transition times through these successive states. From early histological stages (stages I-II), the mean time to acquire cirrhosis is 4 to 6 years4–7; during this period the probability of remaining free of extensive fibrosis is less than 25%.5, 6 However, transplantation-free survival of these patients remains higher than 90%, indicating that death or transplantation are unrealistic endpoints for clinical trials in this population. In contrast, in patients with established cirrhosis, the mean time to acquire a serum bilirubin of 100 μmol/L is about 5 years. The 4-year transplantation-free survival of patients with cirrhosis was estimated to be less than 65% in the combined analysis of ursodeoxycholic acid (UDCA) trials.8 It has been repeatedly shown that bilirubin level and cirrhosis are independent prognostic factors in PBC, the former being the most powerful, even in patients receiving medical therapy (especially UDCA). As a consequence, they may constitute 2 valid surrogate markers to measure the aggravation of PBC and thus the assessment of treatment efficacy.

The potent value of serum bilirubin may be illustrated by the following data drawn from randomized trials. In patients with serum bilirubin levels >35 μmol/L (2 mg/L) the 4-year transplantation free survival is less than 30%; in those with a serum bilirubin >60 μmol/L (3.5mg/L) it is less than 10%.8 These figures illustrate how crucial it is to correctly select a homogenous group of patients to accurately define the power of a study. The combined analysis of UDCA trials comprised 548 patients. Based on the serum bilirubin levels at entry, it was estimated that the 2-year survival of these patients was about 95%, thus requiring 2478 patients to show a statistically significant difference in survival, assuming UDCA prevented 50% of deaths within the 2-year follow-up (P < .05, power 80%).9 As a consequence, the investigators decided to pool all patient data from the 3 main trials and follow them for 4 years to achieve sufficient power to detect an effect of UDCA on delaying death or transplantation. With this data set and duration of follow-up, it was possible to detect a 30% reduction in the risk of dying or being transplanted for patients with moderate to severe disease, a result later confirmed by a combined analysis of trials having the same average duration.10

In addition to their lack of power, a weakness of past clinical trials was the enrollment of a majority of patients with extensive fibrosis or cirrhosis and elevated serum bilirubin levels. The study of sequential histological features of PBC actually show that the majority of bile ducts are irreversibly destroyed when extensive fibrosis occurs, and the activity of the inflammatory process is no longer important when the remodeling process takes place. Thus, it is not unexpected that most anti-inflammatory or immunosuppressive drugs lose their efficacy in advanced-stage disease. Other possible mechanisms may also operate, such as altered metabolism of the drugs (e.g., budesonide, methotrexate, etc.). Another possibility is that cholestasis per se has a major influence on T cell immunity and on several cell-signaling pathways rendering immunosuppressive drugs inadequate in this setting.11–13

The methotrexate story and the present paper from Kaplan and coworkers14 illustrate well the mentioned criticisms and remarks regarding medical therapy in PBC. Following pilot studies showing that hepatic inflammation and bile duct injury improved in selected patients, a first randomized trial reported no benefit from methotrexate.15 This work was criticized mainly because a low dose of methotrexate (7.5 mg/wk) was used. Serum bilirubin levels were not improved by treatment, and progression of fibrosis was not adequately assessed. No difference in major clinical events was observed between the two groups. However, power calculation was not defined and calculation of type II error could not be obtained. More recently, an observational cohort study described the course of 110 patients from the time they began methotrexate until the drug was discontinued or until at least 5 years of continuous use had elapsed.16 As in the Kaplan's study,14 UDCA was added to methotrexate in most patients. Most patients had histological stage II or III at entry into the study. Using Cox regression analysis, the relative risk of failure (death, transplantation) associated with methotrexate use did not approach statistical significance, while the risk of failure for patients on UDCA was significantly lower than for those not on UDCA (RR 0.16, P = .006). A total of 49 paired liver biopsies were obtained in the 57 patients completing 5-year treatment; unfortunately, multistate modeling approach, the most powerful and accurate method to determine the progression toward well-defined stages, was not used. Because methotrexate was not well tolerated and there was no evidence of efficacy, the authors concluded that methotrexate use should not be routinely recommended.

In this issue, Kaplan et al.14 report the results of a small trial of colchicine plus UDCA vs. methotrexate plus UDCA. The study began in 1988. Power calculation was not adequately performed to plan study duration and sample size, despite several sets of data available at this time.4 On average, half of the patients had cirrhosis, and the bilirubin levels ranged from less than 10 μmol/L up to 160 μmol/L, making the study population very heterogeneous. A total of 60 patients were enrolled, and a detailed and precise 10-year follow-up was completed in most of them. As expected, no statistically significant difference in survival-free liver transplantation was observed among the 2 groups. Paired comparison changes in biochemistries were not done among the groups; thus, the results are very difficult to interpret. Nevertheless, two findings are of some interest. First, 29 patients remained in the study; among them, 16 on colchicine and 10 on methotrexate had baseline and 10-year liver biopsies. No patient whose baseline histological stage was I to III progressed to cirrhosis. It is very unlikely that such a result occurred by chance, given the available data on spontaneous histological progression of PBC. Moreover, there was a corresponding improvement in serum bilirubin values in these patients. Responders did not have extensive fibrosis or cirrhosis, and their bilirubin levels were lower than that of the nonresponders. These data suggest a real beneficial effect of methotrexate in a subset of selected patients with early-stage disease.

The third study, published in abstract form17 and called the “PBC ursodiol and methotrexate or placebo study, or PUMPS” randomized patients to UDCA and methotrexate or to UDCA plus placebo. Power calculation was carefully performed, based on the following assumptions: a trial of 260 patients randomized equally to the two treatment arms, an exponential expected survival, and the use of a two-sided log rank test at a significance level of 0.05. Most of the patients (226) entered into the trial with a serum bilirubin of 17 μmol/L or less. About half of the patients had early-histological-stage disease. As expected, treatment failures (death, liver transplantation, variceal bleeding, ascites, a doubling bilirubin to 40 μmol/L or histological progression by 2 stages or to cirrhosis,) were mainly observed in patients with extensive fibrosis or cirrhosis. Assuming a 5-year survival in the UDCA-placebo arm of 80% (a realistic figure), the 5-year survival in the UDCA plus methotrexate arm would need to be at least 90% to achieve a power of 80% within 8 years. After a 7-year follow-up, the hazard ratio for death with or without orthotopic liver transplantation was such that the trial was stopped prematurely, due to futility. Therefore, methotrexate should not be routinely recommended as monotherapy or as adjuvant therapy to UDCA in PBC.

What is the outlook of medical therapy for PBC? As for immunosuppressive drugs, UDCA should be given early in the natural history, i.e. when bile duct inflammation is active and before irreversible bile duct loss, to slow the progression toward cirrhosis and the terminal phase.6 The best and independent predictors of development of cirrhosis are the stage of the disease (according to Ludwig staging), and for any stage, the serum bilirubin and albumin levels and the severity of the lymphocytic periportal piecemeal necrosis. Normalization of biochemistries under UDCA, observed in 30% to 35% of the patients, is surprisingly of less value than the aforementioned criteria; interlobular bile duct destruction and loss is also an important criterion; but it is also of less practical value, probably because of sampling error and variability of quantification. The influence of these prognostic indicators has been described using continuous-time multistate modeling.7 We showed, for example, that a patient with stage II disease, moderate lymphocytic piecemeal necrosis, and a slight elevation of serum bilirubin <34 μmol/L (<2 mg/dL) had a 10-fold higher probability to develop cirrhosis than a patient with stage II disease and serum bilirubin level <17 μmol/L (<1 mg/dL) and a mild lymphocytic piecemeal necrosis. Liver biopsy and serum bilirubin or albumin levels have considerable importance for the management of PBC and design of future trialsA body of evidence suggests that adjuvant therapy, mainly immunosuppressive drugs, especially those having some efficacy in past trials (i.e. corticosteroids and cyclosporin), should target the patients with these characteristics. However, only appropriate well-designed and well-done studies will be able to quantify the possible effect of the combination.


  1. Top of page
  • 1
    Mahl TC, Shockcor W, Boyer JL. Primary biliary cirrhosis: survival of a large cohort of symptomatic and asymptomatic patients followed for 24 years. J Hepatol 1994; 20: 701713.
  • 2
    Prince M, Chetwynd A, Newman W, Metcalf JV, James OFW. Survival and symptom progression in a geographically based cohort of patients with primary biliary cirrhosis: follow-up for up to 28 years. Gastroenterology 2002; 123: 10441051.
  • 3
    Shapiro JM, Smith H, Schaffner F. Serum bilirubin: a prognostic factor in primary biliary cirrhosis. Gut 1979; 20: 137140.
  • 4
    Christensen E, Neuberger J, Crowe JP, Altman DG, Popper H, Portmann B, et al. Beneficial effect of azathioprine and prediction of prognosis in primary biliary cirrhosis. Final results of an international trial. Gastroenterology 1985; 89: 10841091.
  • 5
    Locke IIIGR, Therneau TM, Ludwig J, Dickson ER, Lindor KD. Time course of histological progression in primary biliary cirrhosis. HEPATOLOGY 1996; 23: 5256.
  • 6
    Corpechot C, Carrat F, Bonnand A-M, Poupon RE, Poupon R. The effect of ursodeoxycholic acid therapy on liver fibrosis progression in primary biliary cirrhosis. HEPATOLOGY 2000; 32: 1196-1199.
  • 7
    Corpechot C, Carrat F, Poupon R, Poupon RE. Primary biliary cirrhosis: Incidence and predictive factors of cirrhosis development in ursodiol-treated patients. Gastroenterology 2002; 122: 652-658.
  • 8
    Poupon RE, Lindor KD, Cauch-Dudek K, Dickson ER, Poupon R, Heathcote EJ. Combined analysis of randomized controlled trials of ursodeoxycholic acid in primary biliary cirrhosis. Gastroenterology 1997; 113: 884-890.
  • 9
    Carithers RL. Primary biliary cirrhosis: specific treatment. Clin Liver Dis 2003; 7: 923-939.
  • 10
    Poupon RE. Ursodeoxycholic acid for primary biliary cirrhosis: lessons from the past - issues for the future. J Hepatol 2000; 32: 689-702.
  • 11
    Calmus Y, Weill B, Ozier Y, Chereau C, Houssin D, Poupon R. Immunosuppressive properties of chenodeoxycholic and ursodeoxycholic acids in the mouse. Gastroenterology 1992; 103: 617-621.
  • 12
    Podevin P, Calmus Y, Bonnefils MT, Veyrunes C, Chereau C, Poupon R. Effect of cholestasis and bile acids on interferon-induced 2′,5′-adenylate synthetase and NK cell activities. Gastroenterology 1995; 108: 1192-1198.
  • 13
    Podevin P, Rosmorduc O, Conti F, Calmus Y, Meier PJ, Poupon R. Bile acids modulate the interferon signalling pathway. HEPATOLOGY 1999; 29: 1840-1847.
  • 14
    Kaplan MM, Cheng S, Price LL, Bonis PA. A randomized controlled trial of colchicine plus ursodiol versus methotrexate plus ursodiol in primary biliary cirrhosis: ten-year results. HEPATOLOGY 2004; 39: 915923.
  • 15
    Hendrickse MT, Rigney E, Giaffer MH, Soomro I, Triger DR, Underwood JCE, et al. Low-dose of methotrexate is ineffective in primary biliary cirrhosis: long-term results of a placebo-controlled trial. Gastroenterology 1999; 117: 400407.
  • 16
    Bach N, Bodian C, Bodenheimer HC, Croen E, Berk PD, Thung SN, et al. Methotrexate therapy for primary biliary cirrhosis. Am J Gastroenterol 2003; 98: 187193.
    Direct Link:
  • 17
    Combes B, Emerson SS, Flye NL. The primary biliary cirrhosis (PBC) ursodiol (UDCA) plus methotrexate (MTX) or its placebo study (PUMPS): a multicenter randomized trial [abstract]. HEPATOLOGY 2003; 38 (Suppl 1): 210.