Original Article
A Revised Natural History Model for Primary Sclerosing Cholangitis

https://doi.org/10.4065/75.7.688Get rights and content

Objective

To describe a natural history model for primary sclerosing cholangitis (PSC) that is based on routine clinical findings and test results and eliminates the need for liver biopsy.

Patients and Methods

Using the Cox proportional hazards analysis, we created a survival model based on 405 patients with PSC from 5 clinical centers. Independent validation of the model was undertaken by applying it to 124 patients who were not included in the model creation.

Results

Based on the multivariate analysis of 405 patients, a risk score was defined by the following formula: R =0.03 (age [y]) + 0.54 loge (bilirubin [mg/dL]) + 0.54 loge (aspartate aminotransferase [U/L]) + 1.24 (variceal bleeding [0/1]) – 0.84 (albumin [g/dL]). The risk score was used to obtain survival estimates up to 4 years of follow-up. Application of this model to an independent group of 124 patients showed good correlation between estimated and actual survival.

Conclusions

A new model to estimate patient survival in PSC includes more reproducible variables (age, bilirubin, albumin, aspartate aminotransferase, and history of variceal bleeding), has accuracy comparable to previous models, and obviates the need for a liver biopsy.

Section snippets

Patients

Table 2 describes the 2 sources of data used in this work. The first data were the existing data used in the multicenter model. These data were derived from patients at John Radcliffe Hospital, Oxford, England; King's College, London, England; New England Medical Center, Boston, Mass; Thomas Jefferson University, Philadelphia, Pa; and Mayo Clinic, Rochester, Minn. The second data were generated from a recent randomized trial conducted at Mayo Clinic in which ursodeoxycholic acid was evaluated

Results

In the univariate analysis, the following variables had a significant (P<.01) influence on patient survival (Table 3): age; presence of ascites, hepatomegaly, splenomegaly, and variceal bleeding; hemoglobin level, platelet count, and prothrombin time; and the serum levels of aspartate aminotransferase (AST), albumin, and total bilirubin. Serum alkaline phosphatase level (P=.04), presence of inflammatory bowel disease (P=.16), and male sex (P=.08) showed marginal significance, and presence of

Discussion

Natural history models in PBC and PSC have been used as clinical and research tools.10, 11, 12 In comparison with the PBC model, our previous PSC model has not been widely used. One of the obstacles may have been that the model requires a liver biopsy to assess histologic stage, a notable drawback in application of the model.13 Second, previous models contained variables that were subjective in nature. For example, variables such as splenomegaly or inflammatory bowel disease may be defined

Acknowledgments

We thank Drs Roger W. Chapman (John Radcliffe Hospital, Oxford, England), Marshall M. Kaplan (New England Medical Center, Boston, Mass), Willis Maddrey (Texas Southwestern University, Dallas), and Roger Williams (King's College, London, England) for generously providing data that made this analysis possible.

References (29)

  • KD Lindor et al.

    Urosodiol for primary sclerosing cholangitis

    N Engl Med

    (1997)
  • MD Schluchter

    Methods for the analysis of infomatively censored longitudinal data

    Stat Med

    (1992)
  • JM Robins

    An analytic method for randomized trials with informative censoring; part 1

    Lifetime Data Anal

    (1995)
  • EV Slud et al.

    Dependent competing risks and summary survival curves

    Biometrika

    (1983)
  • Cited by (351)

    • Prognostic models and autoimmune liver diseases

      2023, Best Practice and Research: Clinical Gastroenterology
    • Liver Transplantation: Contraindication and Ineligibility

      2023, Journal of Clinical and Experimental Hepatology
    View all citing articles on Scopus

    This study was supported by grant DK34238 from the National Institutes of Health, Bethesda, Md.

    View full text