Biochemical response to ursodeoxycholic acid and long-term prognosis in primary biliary cirrhosis

Authors

  • Christophe Corpechot,

    Corresponding author
    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    • Service d'Hépatologie, Centre de Référence des Maladies Inflammatoires des Voies Biliaires, Hôpital Saint-Antoine, AP-HP, 184 rue du Faubourg Saint-Antoine, 75571 Cedex 12, Paris, France
    Search for more papers by this author
  • Ludovico Abenavoli,

    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    Current affiliation:
    1. Istituto di Medicina Interna, Università Cattolica del Sacro Cuore, Roma, Italia
    Search for more papers by this author
  • Nabila Rabahi,

    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    Search for more papers by this author
  • Yves Chrétien,

    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    Search for more papers by this author
  • Tony Andréani,

    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    Search for more papers by this author
  • Catherine Johanet,

    1. Immunology Laboratory, Saint-Antoine Hospital, Public Hospitals of Paris (AP-HP), Pierre et Marie Curie University (Paris 6), Paris, France
    Search for more papers by this author
  • Olivier Chazouillères,

    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    Search for more papers by this author
  • Raoul Poupon

    Corresponding author
    1. Hepatology Department, Reference Center for Inflammatory Biliary Diseases, Saint-Antoine Hospital, Public Assistance-Hospitals of Paris (AP-HP), National Institute of Health and Medical Research (INSERM), Mixed Research Unit of Health (UMRS) 680, Pierre et Marie Curie University (Paris 6), Paris, France
    • Service d'Hépatologie, Centre de Référence des Maladies Inflammatoires des Voies Biliaires, Hôpital Saint-Antoine, AP-HP, 184 rue du Faubourg Saint-Antoine, 75571 Cedex 12, Paris, France
    Search for more papers by this author
    • fax: (00) 331 0149282107.


  • Potential conflict of interest: Nothing to report.

Abstract

Biochemical response to ursodeoxycholic acid (UDCA) in patients with primary biliary cirrhosis (PBC) is variable. It has been recently proposed that an alkaline phosphatase (ALP) decline of more than 40% in baseline value or a normal level after 1 year of UDCA treatment (Barcelona criteria) could serve as a good marker of long-term prognosis. Our aim was to define the best efficient set of biochemistries able to identify UDCA-treated patients at risk of death or liver transplantation (LT). The efficiency of several combinations of serum bilirubin, ALP, and aspartate aminotransferase (AST) threshold values to predict outcome was assessed after 1 year of treatment in 292 patients with PBC. Patients showing ALP <3 upper limit of normal (ULN), AST <2 ULN, and bilirubin ≤1 mg/dL after 1 year of UDCA had a 10-year transplant-free survival rate of 90% (95% confidence interval, 81%–95%), compared to 51% (95% confidence interval, 38%–64%) for those who did not (P < 0.001). Patients were less well discriminated by the Barcelona criteria (79% versus 63%). Independent predictive factors of death or LT were baseline serum bilirubin level >1 mg/dL (relative risk [RR], 1.7), histologic stage ≥3 (RR, 1.5), interface hepatitis (RR, 1.9), and the absence of biochemical response (ALP >3 ULN or AST >2 ULN, or bilirubin >1 mg/dL) (RR, 2.3). Antinuclear antibodies against gp210 or Sp100 proteins were associated with death or LT in univariate but not in multivariate analysis. Conclusion: This study defines the best efficient biochemical response to UDCA, which, independent of baseline predictive factors, identifies patients with PBC with a good long-term prognosis. Patients who fail to achieve this response and those with interface hepatitis or advanced histological stage should be targeted for further therapeutic research. (HEPATOLOGY 2008.)

Ursodeoxycholic acid (UDCA) is currently the only drug approved specifically for the treatment of primary biliary cirrhosis (PBC). Long-term treatment with UDCA (13–15 mg/kg/day) in patients with PBC improves biochemical liver tests, delays histological progression, and prolongs survival without liver transplantation (LT).1–5 UDCA monotherapy appears sufficient for many patients as suggested by long-term observational data.6–9 However, the transplant-free survival rate of UDCA-treated patients remains significantly lower than that of an age-matched and sex-matched control population.10 Therefore, there is a continued need for new therapeutic options in PBC, particularly in patients having a suboptimal biochemical response to UDCA and predictive factors of poor outcome.

We have previously shown that serum bilirubin, histological stage, and interface hepatitis were the main predictive factors of UDCA treatment failure in patients with PBC.10–12 However, these prognostic indices have some limitations: (1) hyperbilirubinemia concerns a minority of treated patients; and (2) histological examination of the liver implies an invasive procedure, that is, biopsy, that precludes its repeated use. Biochemical response to UDCA, especially changes in the serum activities of alkaline phosphatase (ALP) and aminotransferases (alanine aminotransferase [ALT] and aspartate aminotransferase [AST]), may appear as a more sensitive and applicable indicator of treatment efficacy. In this regard, a recent study in Barcelona showed that a decrease of more than 40% in the serum activity of ALP, or normal levels of ALP, after 1 year of UDCA treatment was associated with excellent long-term survival in patients with PBC, which was similar to that estimated for a standardized control population.8 By contrast, the survival rate of patients without such criteria of response to UDCA was significantly altered.

In this context, our aim was: (1) to determine the best biochemical criteria of response to UDCA that allowed the accurate prediction of long-term survival of patients with PBC (that is, to define the efficient biochemical response to UDCA); (2) to evaluate the prognostic impact of the Barcelona criteria in a large independent cohort of patients with PBC; and (3) to determine whether the need for histological examination of the liver may be reduced by using biochemical criteria of response to UDCA to assess long-term prognosis in PBC.

Abbreviations

ALP, alkaline phosphatase; ALT, alanine aminotransferase; ANA, antinuclear antibody; AST, aspartate aminotransferase; CI, confidence interval; LT, liver transplantation; ND, nuclear dot; PBC, primary biliary cirrhosis; RR, relative risk; UDCA, ursodeoxycholic acid; ULN, upper limit of normal.

Patients and Methods

Patient Population.

All the patients included in the study had serologically and histologically proven PBC, were treated with UDCA at the daily dose of 13 to 15 mg/kg of body weight and had documented biochemical analyses (that is: serum bilirubin concentration; activities of ALP, gamma-glutamyl transferase, AST, and ALT; serum albumin concentration; and prothrombin index) before and after 1 year of treatment with UDCA. Ineligibility criteria included a diagnosis of autoimmune hepatitis overlap syndrome as previously defined,13 a positive serology for hepatitis B virus or hepatitis C virus, the use of corticosteroids or immunosuppressive drugs during a duration of more than 2 months, and a follow-up of less than 1 year.

The histological stage was defined according to the Ludwig classification.14 Interface hepatitis was graded as follows: absent or mild, occasional foci of piecemeal necrosis at the periphery of less than half of the portal tracts; moderate, segmental necrosis at the periphery of more than half of the portal tracts or circumferential necrosis in less than half of the portal tracts; or severe, circumferential necrosis in almost all portal tracts. A minimum of 10 portal tracts was required for the estimation of ductopenia. The degree of ductopenia was calculated as the number of portal tracts without interlobular bile ducts divided by the total number of portal tracts. Interlobular bile duct paucity was defined as a ductopenia of more than 50%.

Serum antinuclear antibodies (ANAs) were consistently determined at diagnosis by indirect immunofluorescence on rat tissue sections and Hep-2 cells. Sera were considered positive at a dilution of ≥1:80. Sera giving a rim-like/membranous pattern were tested for anti-gp210 antibodies. Those giving a multiple nuclear dots (ND) pattern were tested for anti-Sp100 antibodies. The presence of anti-gp210 antibodies was determined by a standardized enzyme-linked immunosorbent assay using a specific synthetic polypeptide as described.15 The detection of anti-Sp100 antibodies was carried out using a commercially available enzyme-linked immunosorbent assay kit (INOVA Diagnostics Inc., San Diego, CA). Data on ANAs were not available in a subgroup of patients who had participated to a multicenter placebo-controlled trial of UDCA at the beginning of the 1990s.1

Data Analysis.

Data were expressed as mean ± standard deviation. A P value <0.05 was considered statistically significant. The comparisons between biochemical variables before and after 1 year of UDCA treatment were performed using the Student t test for paired data. The endpoint for treatment efficacy was survival without LT. The rates of survival were estimated by the Kaplan-Meier method. Patient survival was compared with two different survival estimates: (1) the survival rate expected in a standardized cohort of French subjects matched for age, sex, and follow-up period; and (2) the spontaneous survival predicted by the updated Mayo model for the first 7-year period. These estimates were calculated as described.10 The comparisons between observed and predicted survivals and the calculation of the RRs of death or LT were carried out using the log-rank test. Univariate and multivariate associations between the prognostic factors and the risk of death or LT were determined using Cox regression analysis. For each quantitative variable, a cutoff value was defined to split the data into two categories and the RRs with 95% confidence intervals (CIs) were estimated.

Several definitions of the biochemical response to UDCA were studied, including the Barcelona definition.8 Three biochemical variables were selected for their strong association with death or LT in Cox analysis: serum bilirubin, ALP, and AST. The predictive significance of these variables was estimated for several combinations of threshold values at 1 year of UDCA treatment. The bilirubin level was categorized as ≤1.0 mg/dL or >1.0 mg/dL. ALP activity was categorized as: ≤1.5-fold the upper limit of normal (ULN); >1.5 ULN and ≤2.0 ULN; >2.0 ULN and ≤3.0 ULN; or >3.0 ULN. AST activity was categorized as: ≤1.0 ULN; >1.0 ULN and ≤1.5 ULN; >1.5 ULN and ≤2.0 ULN; or >2.0 ULN. The likelihood ratio test was used for the selection of the best combination of these cutoff values in terms of survival prediction.

Results

A total of 292 patients complied with the inclusion criteria. A total of 136 patients (47%) had been previously enrolled in the 2-year French multicentric randomized controlled UDCA trials with an open follow-up extension to 4 years, and 156 (53%) were followed up between 1990 and 2006 under similar conditions in the Saint-Antoine Hospital, Paris.

The characteristics of the patients at enrollment and after 1 year of UDCA therapy are shown in Table 1. At baseline, more than half of the patients (57%) were in an early stage of the disease (that is, histological stage I-II). Cirrhosis was present in only 17% of cases. After 1 year of UDCA treatment, the serum activities of ALP, gamma-glutamyl transferase, AST, and ALT were decreased by about 50% (P < 0.0001), and the serum concentration of bilirubin was decreased by 25% (P < 0.01), as compared to baseline values.

Table 1. Characteristics of Patients (n = 292) at Enrollment and After 1 Year of UDCA Treatment
CharacteristicAt EnrollmentAfter 1 Year
  • Quantitative data are expressed as mean ± standard deviation (SD).

  • *

    P < 0.01;

  • **

    P < 0.0001. GGT, gamma-glutamyl transferase.

Age (years)53 ± 12
Female gender92%
Total serum bilirubin (mg/dL)1.4 ± 2.21.0 ± 1.1*
ALP (ULN)4.0 ± 3.22.2 ± 1.8**
GGT (ULN)10.8 ± 10.04.6 ± 5.2**
AST (ULN)2.1 ± 1.61.4 ± 1.4**
ALT (ULN)2.4 ± 1.91.5 ± 1.4**
Serum albumin (g/L)40 ± 5
Prothrombin index (%)96 ± 10
Mayo risk score4.5 ± 1.2
Histological stage 3—443%

The patients were treated for a mean period of 6.1 ± 4.3 years (median, 5.3 years; range, 1.0–21.5 years). A total of 53 patients (18%) died (n = 28) or underwent LT (n = 25) during the study period. Nineteen deaths (68%) were liver-related. The overall survival was 94% at 5 years and 85% at 10 years (Fig. 1). Despite a trend for a lower survival, the difference with the standardized control population did not reach the level of significance (P = 0.11; RR, 1.6; 95% CI, 0.9–2.7). Survival without LT was significantly lower than that of the control population (P < 0.0001; RR, 2.6; 95% CI, 1.6–4.0) but remained higher than the survival predicted by the updated Mayo model (P = 0.01; RR, 0.6; 95% CI, 0.4–0.9).

Figure 1.

Survival with and without transplantation of the overall series of UDCA-treated patients with PBC (n = 292). The thin curve represents survival with transplantation. The thick curve represents survival without transplantation. The survival of the control population (thin dotted curve) and the survival predicted by the updated Mayo model (thick dotted curve) are indicated. The overall survival was not different from that of the control population, despite an excess of deaths (P = 0.11). Survival without transplantation was significantly lower than that of the control population (P < 0.0001) and higher than the survival predicted by the updated Mayo model (P = 0.01).

Several definitions of the biochemical response to UDCA were tested (Table 2). The most accurate discrimination of the patients according to the risk of death or LT was given by the following definition: serum bilirubin ≤1 mg/dL, ALP ≤3 ULN, and AST ≤2 ULN. A total of 179 patients (responders, 61%) responded to UDCA according to this definition. This subgroup of patients showed fewer severe clinical, biochemical, and histological features at baseline than nonresponders (Table 3). They were also less frequently positive for anti-gp210 or anti-Sp100 antibodies (17%) than nonresponders (32%; P < 0.05).

Table 2. Selection of the Most Discriminative Definition of the 1-Year Biochemical Response to UDCA for the Prediction of Death or Liver Transplantation
Biochemical Response at 1 YearLikelihood Ratio TestRisk Ratio (95% CI)PResponse Rate (%)
  • *

    Barcelona criteria.

Bilirubin ≤1 mg/dL, ALP ≤1.5 ULN, and AST ≤1 ULN17.40.45 (0.27–0.68)<0.000132
Bilirubin ≤1 mg/dL, ALP ≤2 ULN, and AST ≤1 ULN21.50.44 (0.27–0.64)<0.000139
Bilirubin ≤1 mg/dL, ALP ≤1.5 ULN, and AST ≤1.5 ULN22.60.43 (0.27–0.63)<0.000137
Bilirubin ≤1 mg/dL, ALP ≤2 ULN, and AST ≤1.5 ULN34.40.38 (0.24–0.55)<0.000148
Bilirubin ≤1 mg/dL, ALP ≤2 ULN, and AST ≤2 ULN40.10.37 (0.24–0.52)<0.000152
Bilirubin ≤1 mg/dL, ALP ≤3 ULN, and AST ≤2 ULN48.90.35 (0.25–0.48)<0.000161
Decrease in ALP >40% of baseline value or normal level*8.80.66 (0.51–0.87)<0.00165
Table 3. Baseline Characteristics and Outcome of Patients According to the 1-Year Biochemical Response to UDCA
CharacteristicResponders* (n = 179)Nonresponders (n = 113)P
  • *

    Patients with total serum bilirubin <1 mg/dL, ALP ≤ 3 ULN, and AST ≤2 ULN after 1 year of UDCA.

  • ANAs status was available in 133 (46%) patients.

Age (years)53 ± 1153 ± 13NS
Female gender93%91%NS
Duration of treatment (years)6.6 ± 4.45.2 ± 4.1<0.001
Pruritus27%53%<0.001
Total serum bilirubin (mg/dL)0.9 ± 2.02.1 ± 2.3<0.0001
ALP (ULN)3.0 ± 2.05.9 ± 3.9<0.0001
GGT (ULN)8.9 ± 7.714.1 ± 12.5<0.0001
AST (ULN)1.7 ± 1.32.9 ± 1.8<0.0001
ALT (ULN)2.1 ± 1.63.1 ± 2.2<0.0001
Serum albumin (g/L)41.0 ± 4.537.1 ± 5.4<0.0001
Prothrombin index (%)98 ± 893 ± 13<0.001
Mayo risk score4.1 ± 0.95.2 ± 1.3<0.0001
Anti-gp210/Sp100 antibodies17%32%<0.05
Histological stage 3–432%64%<0.0001
Interface hepatitis42%62%<0.005
Ductopenia56%84%<0.0001
Treatment failure12 (7%)41 (36%)<0.0001
 Death919 
 Liver-related death514 
 Liver transplantation322 

Treatment failure (death or LT) was recorded in 12 responders (7%) and 41 nonresponders (36%). The risk of death or LT was decreased by 60% in responders as compared to nonresponders (P < 0.0001; RR, 0.4; 95% CI, 0.3–0.5). The 10-year rate of survival without LT was 90% (95% CI, 81%–95%) in responders and 51% (95% CI, 38%–64%) in nonresponders (Fig. 2). The transplant-free survival rate of responders was similar to that estimated for the standardized control population (P = 0.8). The transplant-free survival rate of nonresponders was not different from the survival rate predicted by the updated Mayo model (P = 0.9).

Figure 2.

Survival without transplantation, according to the 1-year biochemical response to UDCA. The thin curve represents survival of responders (n = 179). The thick curve represents survival of nonresponders (n = 113). The survival of the control population (thin dotted curve) and the survival predicted for nonresponders by the updated Mayo model (thick dotted curve) are indicated. The transplant-free survival of responders was significantly higher than that of nonresponders (P < 0.0001; RR, 0.4; 95% CI, 0.3–0.5) and similar to that of the control population (P = 0.8). The transplant-free survival of nonresponders was not different from that predicted by the updated Mayo model (P = 0.9).

In contrast, 189 patients (65%) responded to UDCA, according to the Barcelona criteria. Treatment failure was recorded in 29 responders (15%) and 24 nonresponders (23%). The risk of death or LT was decreased by 30% in responders, compared to nonresponders (P < 0.01; RR, 0.7; 95% CI, 0.5–0.9). The 10-year rate of survival without LT was 79% (95% CI, 70%–86%) in responders and 63% (95% CI, 47%–74%) in nonresponders (Fig. 3). The transplant-free survival rate of responders remained significantly lower than that of the control population (P < 0.01; RR, 2.4; 95% CI, 1.4–4.1). The transplant-free survival rate of nonresponders was not different from the survival rate predicted by the updated Mayo model (P = 0.4).

Figure 3.

Survival without transplantation, according to the Barcelona criteria of biochemical response to UDCA. The thin curve represents survival of responders (n = 189). The thick curve represents survival of nonresponders (n = 103). The survival of the control population (thin dotted curve) and the survival predicted for nonresponders by the updated Mayo model (thick dotted curve) are shown. The transplant-free survival of responders was significantly higher than that of nonresponders (P < 0.01; RR, 0.7; 95% CI, 0.5–0.9), but remained lower than that of the control population (P < 0.01; RR, 2.4; 95% CI, 1.4–4.1). The transplant-free survival of nonresponders was not different from that predicted by the updated Mayo model (P = 0.4).

Clinical, biochemical, and histological data were available in all patients. ANA status was available in 133 (46%) of them. The prognostic factors influencing survival without LT in univariate analysis are shown in Table 4. Risk of death or LT was increased in patients with older age; higher baseline bilirubin, ALP, or AST levels; lower albuminemia or prothrombin index; higher Mayo risk score; anti-gp210 or anti-Sp100 antibodies (that is, PBC-specific ANAs); advanced histological stage (that is, III-IV); ductopenia; interface hepatitis; and incomplete biochemical response to UDCA. In multivariate analysis, four variables were retained as independent predictive factors for death or LT: baseline serum bilirubin level, histological stage, severity of interface hepatitis, and biochemical response to UDCA (Table 5). The same independent predictors were obtained when non-liver-related deaths were excluded from the analysis. When cirrhotic patients were excluded at baseline, biochemical response, interface hepatitis, and baseline serum bilirubin concentration remained independently linked to death or LT.

Table 4. Univariate Analysis of Potential Predictive Factors for Death or Liver Transplantation in UDCA-Treated Patients
VariablePRelative Risk (95% CI)
  • *

    At the onset of treatment.

  • Total serum bilirubin >1 mg/dL, or ALP >3 ULN, or AST >2 ULN.

Age >55 years*0.01611.3 (1.0–1.6)
Total serum bilirubin >1 mg/dL*<0.00012.7 (2.0–3.7)
ALP > 3ULN*0.00391.6 (1.2–2.0)
AST > 2ULN*0.01381.6 (1.3–2.1)
Serum albumin <38 g/L*<0.00012.0 (1.6–2.7)
Prothrombin index <80%*<0.00012.2 (1.7–2.9)
Mayo score >4.5*<0.00012.5 (1.8–3.7)
Anti-gp210/Sp100 antibodies*0.04011.5 (1.0–2.1)
Histological stage 3–4*<0.00012.5 (1.9–3.2)
Moderate to severe interface hepatitis*<0.00012.0 (1.5–2.7)
Ductopenia*0.01261.8 (1.3–2.7)
Absence of biochemical response at 1 year according to the Barcelona criteria0.00151.5 (1.2–2.0)
Absence of biochemical response at 1 year according to the present study's criteria<0.00012.9 (2.1–4.2)
Table 5. Multivariate Analysis of Predictive Factors for Death or Liver Transplantation in UDCA-Treated Patients
VariablePRelative Risk (95% CI)
  • *

    At the onset of treatment.

  • Total serum bilirubin >1 mg/dL, or ALP >3 ULN, or AST >2 ULN.

Total serum bilirubin >1 mg/dL*0.01311.7 (1.1–2.6)
Histological stage 3–4*0.04451.5 (1.0–2.2)
Moderate to severe interface hepatitis*0.00221.9 (1.2–2.9)
Absence of biochemical response at 1 year according to the present study's criteria<0.00012.3 (1.5–3.7)

Discussion

First, the results of this study show that the long-term survival rate for many UDCA-treated patients with PBC is only slightly decreased compared to that estimated for an age-matched and sex-matched control population. This finding is in keeping with our previous data and those of several European cohort studies of long-term treated patients.6, 7, 9 It suggests that the long-term administration of UDCA alone at the recommended daily dose of 13–15 mg/kg is sufficient for many PBC patients. However, the fact that transplant-free survival is still lower than that of the control population strongly indicates that additional therapies are required in some patients.

Second, our data confirm a significant increase in the transplant-free survival of UDCA-treated patients as compared to that predicted by the updated Mayo model. This observation, which has now been reported in five large independent U.S. and European series,6–9, 16 supports a long-term beneficial effect of UDCA in PBC. The comparison of the observed data with the predictions of the Mayo model suggests a 40% decrease in the risk of death or LT under UDCA over a 7-year period. Although these data cannot replace the results of a randomized placebo-controlled trial, they support the conclusions of two independent meta-analyses and of one combined analysis of long-term follow-up trials estimating the reduction in the risk of death or LT under UDCA at between 24% and 32%.4, 17, 18

Third, we show that simultaneous determination of serum bilirubin, ALP, and AST 1 year after the onset of UDCA therapy is predictive of death or LT in PBC. Death or LT were 2.5-fold more frequent in patients who showed either ALP >3 ULN, AST >2 ULN, or serum bilirubin >1 mg/dL at 1 year of treatment. This subgroup of patients with a high risk of premature death or LT accounts for 39% of all patients and is characterized by a 10-year rate of survival without LT of about 50%. In contrast, the survival of the patients who showed none of these biochemical indices was similar to that of the control population.

In our cohort, the patients who experienced an ALP decrease >40%, or normal levels, after 1 year of UDCA (that is, the Barcelona criteria) still maintained a decreased survival rate as compared to the control population. This observation, which contrasts with the findings of the Barcelona study, may result from a higher proportion of cirrhosis at baseline in our study (17% versus 9%). However, the exclusion of the patients with cirrhosis at baseline did not improve the predictive significance of the Barcelona criteria as compared to ours. This result strongly suggests that the decrease in serum ALP is not sufficient by itself to accurately predict the long-term outcome of patients with PBC under UDCA, and that the selection of UDCA-treated patients for future clinical trials should be based not only on serum ALP but also on AST and bilirubin levels. This finding is not really unexpected because serum bilirubin is one of the strongest surrogate markers of PBC prognosis, even under UDCA,19, 20 and AST activity has been shown to be associated with the severity of both fibrosis and interface hepatitis in UDCA-treated patients.21 These data, however, need to be confirmed by external validation from independent cohorts of patients.

As expected, poor biochemical response was more frequently observed in patients with the most severe biochemical and histological features at baseline. However, it should be noted that biochemical response was independent of baseline serum bilirubin and histological stage for the prediction of survival. For example, about 40% of patients with advanced histological stage or hyperbilirubinemia at baseline experienced a good biochemical response to UDCA, whereas 15% of patients who did not respond well had neither advanced histological stage nor hyperbilirubinemia at baseline. These findings indicate that biochemical response to UDCA is an essential predictor of long-term outcome, even in patients lacking bad prognostic factors at baseline.

We failed to find any increase in the survival rate of patients with poor biochemical response as compared to the predicted natural survival rate. This finding raises the question as to whether UDCA should be stopped (switching to an alternative drug) or continued (step-up combination therapy) in this situation. Clinical trials aimed at comparing these two therapeutic strategies in nonresponders to UDCA should be conducted. It must be underlined that the predictions of the Mayo model are limited to the first 7 years of follow-up and therefore are not appropriate to test a small but possibly significant effect of UDCA on more long-term survival.

We and others have previously shown that PBC-specific ANAs directed against antigens of the nuclear pore complex or ND may be associated with features of more severe disease and poor outcome.22–25 However, this finding needs to be validated in larger longitudinal studies. Sp100 and gp210 proteins are the main antigenic targets of anti-ND and anti-nuclear pore complex antibodies, respectively.26 Herein, we confirm that the presence of anti-gp210 or anti-Sp100 antibodies in the serum is more frequent in patients with poor response to UDCA and is associated with an increased risk of death or LT. However, multivariate analysis did not retain this parameter as an independent prognostic factor. As our study was not specifically designed to investigate the prognostic role of these antibodies (ANA status was available in only half of our patients), a lack of statistical weight is probable. Because all the patients were positive for antimitochondrial antibodies, it was not possible to assess the potential influence of antimitochondrial antibody–negative cases on prognosis.

We have previously shown that the severity of interface hepatitis was associated with histological progression and cirrhosis development in UDCA-treated patients with the classical form of PBC.12 Other authors have more recently found that this histological feature was associated with an increased risk of poor biochemical response to UDCA.27 We show here that interface hepatitis, in the absence of additional features of autoimmune hepatitis, is able to influence the survival of patients with PBC, independent of their biochemical response to UDCA. In a previous study, we showed that AST level was the only biochemical marker linked to the severity of interface hepatitis in PBC.21 However, this parameter was of insufficient diagnostic performance to be used as a reliable surrogate marker of interface hepatitis. Consequently, our data indicate that a liver biopsy examination is still required for an accurate evaluation of PBC prognosis.

In summary, we show that the combined evaluation of serum ALP, AST, and bilirubin after a 1-year period of UDCA therapy provides an accurate estimation of the long-term risk of death or LT in PBC and should facilitate the selection of patients for future clinical trials. The simple biochemical combination (ALP >3 ULN, or AST >2 ULN, or bilirubin >1 mg/dL) we propose here is likely to become a useful tool in the everyday care of patients with PBC, whereas the decreased percentage in serum ALP alone seems less efficient to predict the long-term efficacy of UDCA. Finally, we show that the severity of interface hepatitis impacts the survival of patients with PBC in an independent manner, a result that underlines the crucial role of liver biopsy in the screening of patients with PBC, especially when selecting potential candidates for additional anti-inflammatory therapies.

Ancillary