The natural history of primary biliary cirrhosis is extremely long and may extend over decades. An important paper [2] recently showed that patients who are positive for antimitochondrial antibody but have normal liver biochemical results will, with time, frequently develop clinically obvious primary biliary cirrhosis, thereby further extending the natural history of this disease backward in time. The disease is usually one of slow clinical progression. Most patients eventually develop such symptoms as pruritus or nonspecific fatigue. Overall patient survival is decreased, particularly when the disease is untreated. Various predictive models have been developed to help better assess patient prognosis; most of these models use bilirubin as a major component with or without histologic staging [3-5].

The histology of the disease is usually grouped into four categories that reflect the extent of inflammation within or away from the portal space, the degree of fibrosis, and the presence of cirrhosis [6]. The histologic progression through these stages is slow, averaging 1.5 years per stage [7]. The florid duct lesion is essentially pathognomonic for primary biliary cirrhosis but is seen in less than 10% of patients [8]. It is interesting that Kaplan and colleagues found florid duct lesions in two of the three patients with biochemical normalization and histologic improvement. It is also important to note that sampling variation is definitely seen in liver biopsy samples of patients with primary biliary cirrhosis. In one study [7], we found evidence of histologic regression in 8% of biopsy specimens; we interpreted this regression as an indicator of sampling variation. Other researchers [9, 10] have estimated higher degrees of sampling variation in cholestatic liver disease. The issue of sampling variation must be carefully considered in interpreting Kaplan and colleagues' results.

Several therapeutic options have been studied and are now available to patients with primary biliary cirrhosis. Ursodeoxycholic acid is the drug most commonly used to treat this condition, but it has not yet been approved by the U.S. Food and Drug Administration. Ursodeoxycholic acid is extremely safe and well tolerated and has been shown to decrease clinical treatment failure and improve transplant-free survival [11-13]. As Kaplan and colleagues have indicated, the effects of ursodeoxycholic acid on liver histologic findings are quite uncertain. Colchicine is used frequently, and biochemical improvement has been seen in most studies with this drug [14]. The effects of colchicine on survival are uncertain, and no evidence of efficacy of histologic improvement has been noted [14, 15]. Kaplan and colleagues introduced methotrexate for the treatment of primary biliary cirrhosis, and the drug remains under investigation. Early clinical trials using this drug in conjunction with ursodeoxycholic acid did not show obvious benefit compared with ursodeoxycholic acid alone [16]. A large-scale multicenter randomized trial is currently being conducted in the United States to address just this issue. An important difference in the approach described by Kaplan and colleagues is the use of methotrexate in patients who had been unresponsive to other therapy. The notion of a stepped approach to therapy is frequently used in other fields of medicine but seldom has been applied in patients with liver disease, probably because of the limited repertoire of effective drugs available for patients with chronic liver diseases such as primary biliary cirrhosis. Certainly, however, this approach seems logical and deserves further evaluation.

Kaplan and coworkers also raise issues about the appropriateness of various end points in future trials of patients with primary biliary cirrhosis. Such symptoms as pruritus are difficult to quantitate and may fluctuate. Fatigue is especially difficult to measure; is nonspecific; and, in many instances, may not be at all related to the underlying liver disease. Alterations in serum biochemical results are easy to measure; unfortunately, changes in liver-related biochemical variables may not be reliable predictors of improvement in the clinical course. Other outcomes, such as death or liver transplantation, obviously can be considered hard end points in more ways than one. However, studies that use such end points for patients with primary biliary cirrhosis need to include a very large number of patients. In addition, follow-up needs to be extensive, particularly if patients with early disease (for example, those who might be most appropriately treated with newer agents, such as methotrexate) are studied. Liver transplantation as an end point has often been considered to be a surrogate for death. The exact timing of liver transplantation can vary from center to center, however, and some have therefore argued that this is an imprecise end point.

The study by Kaplan and colleagues also raises the issue of histology as a potential clinical end point. Liver histologic results can usually be readily obtained with reasonable safety in patients with primary biliary cirrhosis [17]. However, issues of sampling variability, the effects of the drug on established fibrosis (which forms the basis for much of the staging criteria), and patient acceptance of biopsy create many uncertainties about this measure. If, as has been suggested, specific features of the liver biopsy specimen could really be used to measure treatment efficacy, this histologic approach might provide a reasonable end point about future trials; the latter are a source of clinically meaningful information that is more rapidly attainable than information obtained from death or liver transplantation.

Kaplan and associates' paper thus illustrates many substantial problems in designing trials for patients with primary biliary cirrhosis; these problems include the long natural history of the condition and the difficulty of finding adequate clinical end points. Complicating matters further, the authors' results raise the intriguing possibility that primary biliary cirrhosis may actually include several subtypes that differ as to cause, pathophysiology, and responsiveness to therapy.

Efforts to categorize patients' clinical courses or responses to therapy according to subtype on the basis of usual biochemical findings, the presence of absence or associated disease, or titers of immunoserologic markers have thus far been unrevealing. Clues might be found through the use of a more detailed assessment of liver histologic features. In this regard, the finding of Kaplan and colleagues that two of the three patients in whom complete responses were obtained had florid duct lesions is suggestive.

Further evaluation is needed for the many important methodologic issues that this study raises, including the use of liver histology for subtyping patients; evaluation of a stepped approach to therapy (that is, adding drugs to other therapeutic agents according to initial response); and, finally, the possibility that primary biliary cirrhosis may actually consist of many disease variants that differ in biologically and clinically important ways. Whatever answers emerge from future research into primary biliary cirrhosis, the work of Kaplan and colleagues has already pointed the way to potentially exciting developments in our understanding of the disease and its management.