ORIGINAL ARTICLE
Predictive Scores in Primary Biliary Cirrhosis A Retrospective Single Center Analysis of 204 Patients Arndt Weinmann, MD, MBA,*w Thomas Sattler, MD,z Hans-Peter Unold, MD,y Annette Grambihler, MD,w Andreas Teufel, MD, PhD,8 Sandra Koch,*w Marcus Schuchmann, MD, PhD,z Stefan Biesterfeld, MD, PhD,# Marcus A. Wo¨rns, MD, PhD,*w Peter R. Galle, MD, PhD,w and Henning Schulze-Bergkamen, MD, PhD**
Goals: The aim of this study was to assess the long-term outcome of primary biliary cirrhosis (PBC) patients and to test the clinical value of various outcome models, such as the Mayo Risk Score (MRS), in a large single-center cohort in Germany. Background: PBC is a chronic autoimmune liver disease with a female gender predominance and a peak incidence in the fifth decade of life. PBC is characterized by portal inflammation and immunemediated destruction of intrahepatic bile ducts in liver histology and the presence of antimitochondrial antibodies in the serum of nearly 95% of patients. In 5% to 20% of patients an overlap syndrome with autoimmune hepatitis (AIH) is diagnosed. Ursodeoxycholic acid is widely accepted as the standard medical treatment. Study: A total of 204 patients with PBC or PBC/AIH were retrospectively analyzed with regard to their clinical, biochemical, serological, and histologic features. PBC was diagnosed on the basis of the American Association for the Study of Liver Diseases criteria. Specific PBC scores, such as the MRS, the European and the Yale model, as well as nonspecific scores such as the ChildPugh, the Model for End-stage Liver Disease, and Aspartate Aminotransferase to Platelet Ratio Index score were analyzed for their utility to predict the clinical outcome of patients. Results: One hundred eighty-four patients with PBC alone and 20 with primary biliary cirrhosis/autoimmune hepatitis overlap were followed up for an average of 7.0 (range, 0.5 to 33.2) years. Importantly, baseline values of serum bilirubin, alkaline phosphatase, immunoglobulin M (IgM) and IgG, as well as antimitochondrial antibodies titers did not allow in properly predicting patient’s outcome. The MRS proved clinical applicability. Patients Received for publication November 14, 2013; accepted May 22, 2014. From the *Clinical Registry Unit (CRU); wDepartment of Internal Medicine I, University Medical Center of the Johannes Gutenberg University Mainz, Mainz; zDepartment of Surgery; **Department of Medical Oncology, National Center for Tumor Diseases (NCT), University Clinic of Heidelberg, Heidelberg; yDepartment of Surgery, Kreuznacher Diakonie Hospital, Bad Kreuznach; 8Department of Internal Medicine I, University Clinic of Regensburg, Regensburg; zDepartment of Internal Medicine, Hospital Konstanz, Konstanz; and #Department of Pathology, University Clinic of Du¨sseldorf, Du¨sseldorf, Germany. A.W. and T.S. contributed equally. A.W., T.S., H.-P.U., H.S.-B.: manuscript preparation and data and statistical analysis; T.S., H.-P.U., A.G., A.T., S.B., M.S., and M.A.W.: data collection; A.W., S.K., and M.A.W.: clinical registry database; A.T., M.S., S.B., M.A.W., and P.R.G.: revision of manuscript; and P.R.G.: study supervision. P.R.G. has received lecture fees from the Dr Falk Pharma GmbH. The remaining authors declare that they have nothing to disclose. Reprints: Arndt Weinmann, MD, MBA, Department of Internal Medicine I, Johannes Gutenberg-University Mainz, Langenbeckstrae 1, Mainz 55101, Germany (e-mail: [email protected]). Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved.
438 | www.jcge.com
with an R-value 40% from baseline after 1 year of UDCA treatment was associated with improved prognosis.21 An overlap syndrome (OS) with autoimmune hepatitis (AIH) has been described in 4.8% to 19% of PBC patients.22–31 Patients present with symptoms like fatigue, pruritus, jaundice, arthralgia, and myalgia. Primary biliary cirrhosis/autoimmune hepatitis (PBC/AIH), like PBC itself, is more common in women. Identification of patients with OS is clinically relevant as these patients have a poorer outcome and often respond to immunosuppressive therapy.23,25 In the absence of standardized diagnostic criteria, Chazouille`res and colleagues introduced a score for diagnosis of PBC/AIH.23 This retrospective single center study analyzed comprehensive clinical, serological, and histologic data of 204 patients with PBC and PBC/AIH undergoing UDCA treatment with a median observation period of 7 years. Various prognostic models and scores were applied to predict patient outcome and to test their clinical applicability.
MATERIALS AND METHODS Patient Population All consecutive patients with confirmed PBC with sufficient clinical data that were treated in the University Medical Center of the Johannes Gutenberg University Mainz between January 1990 and December 2009 were included in this retrospective analysis. The center is a quaternary referral center for autoimmune and other chronic liver diseases, treating more advanced stages of liver diseases and also conducting LT. Patients who fulfilled the criteria established by the American Association for the Study of Liver Diseases19 were included in this study. Study entry was defined as the date of the first presentation in our department. No formal ethics approval was required for this strictly retrospective study as was ruled by the local ethics committee (EthikKommission der Landesa¨rztekammer Mainz). Data were stored and analyzed in a specialized clinical registry database developed for this study. Access was strictly limited to the study investigators. Diagnosis of PBC required at least 2 of the 3 following criteria: (1) biochemical evidence of cholestasis [elevated levels of ALP and g-glutamyl transferase (gGT) over a period of 6 mo], (2) serological presence of AMA, and (3) a liver biopsy compatible with PBC or PBC/AIH. Measurement of AMA was performed using indirect immunofluorescence on rat tissue, whereas AMA subtyping was performed using enzyme-linked immunosorbent assay. An overlap-syndrome of PBC with AIH was diagnosed, if both criteria for PBC and AIH were fulfilled. AIH criteria included elevated immunoglobulin g (IgG) levels, serological presence of antinuclear antibodies (ANA), smooth muscle antibodies or soluble liver antigen, a liver biopsy compatible with AIH or PBC/AIH, as well as the exclusion of chronic viral hepatitis.32 Patients underwent regular physical examinations, laboratory assessments [eg, serum bilirubin, serum albumin, ALP, gGT, aspartate aminotransferase (AST), and alanine aminotransferase (ALT)], liver ultrasound, and upper Copyright
r
Predictive Scores in PBC
digestive endoscopy in cases of portal hypertension. Some patients received second liver biopsy to assess progression of liver disease. The histologic stage was assigned according to the Ludwig classification, and early-stage disease was defined as portal or periportal hepatitis in the absence of fibrotic septa or cirrhosis.33 Fibrosis and inflammation levels were classified according to Desmet et al34 and Scheuer.35 UDCA was prescribed with a daily dose of 13 to 15 mg/kg body weight, if advanced liver cirrhosis and pregnancy were excluded. Survival was analyzed as LT-free survival.
Prognostic Scores and Models Child-Pugh and the Model for End-stage Liver Disease (MELD) score were assessed as described.36–38 The Aspartate Aminotransferase to Platelet Ratio Index (APRI) was developed to assess the extent of liver fibrosis in viral hepatitis. It was calculated as following39: APRI = [glutamate oxaloacetate transaminase (GOT)] [upper limit of normal (ULN)] 100)/platelet count [109/L]. The Mayo Risk Score (MRS) was calculated as described40: R-value = 0.871 ln[serum bilirubin (mg/ dL)] 2.53 ln[albumin (mg/dL)] + 0.039 age + 6.843 + 1.02 ln(International Normalized Ratio) + 0.859 ascites (no = 0, yes = 1). The European model was measured as following41: 2.53 (log(serum bilirubin [mmol/dL]) 1.53) + 1.39 ascites (no = 0, yes = 1) 0.085 [albumin (g/L) 34.3] + 0.04 (age–55) + 0.65 upper gastrointestinal bleeding (no = 0, yes = 1). The Yale model was calculated42: 0.037 1.037(age) + 0.74 2.1 hepatomegaly (no = 0, yes = 1) + 0.82 2.26 [serum bilirubinZ5 mg/dL (no = 0, yes = 1)] 0.73 0.48 [serum bilirubin 40% or normalization of ALP after 1 year of UDCA therapy and to Corpechot et al45 (ALP < 3 ULN, GOT 1ULN [n(%)] > 2ULN [n(%)] > 3ULN [n(%)] GGT (ULN) GOT (ULN) > 1ULN [n(%)] > 2ULN [n(%)] GPT (ULN) GLDH (ULN) CHE (LLN) Platelets (LLN) IgM (ULN) > 1ULN [n(%)] IgG (ULN) Serological findings AMA positive [n(%)] AMA/M2 positive [n(%)] ANA positive [n(%)] SMA-positive [n(%)] SLA positive [n(%)] Anti-HAV positive [n(%)] HBsAg positive [n(%)] Anti-HCV positive [n(%)] Histologic findings Ludwig score (I-II/III-IV) Fibrosis (0-2/3-4) Inflammation (0-2/3-4) Other features Splenomegaly [n(%)]
PBC/AIH (n = 20)
95% CI*
P
Pcw
55 (26-71) 18 (90) 8.5 (2.3-32.0)
( 7; 5) (0.21; 4.48) ( 1.0; 5.8)
N.S. N.S. N.S.
N.S. N.S. N.S.
0.50 (0.15-12.08) 23/142 (16) 12/142 (8) 1.24 (0.79-1.56) 10/78 (13) 1.46 (0.28-7.28) 104/145 (72) 52/145 (36) 30/145 (21) 4.79 (0.14-36.11) 1.23 (0.25-23.78) 85/142 (60) 31/142 (22) 1.42 (0.19-25.29) 2.47 (0.24-13.84) 2.02 (0.02-3.95) 1.73 (0.25-4.07) 1.42 (0.27-9.79) 102/136 (75) 0.84 (0.37-2.05)
0.75 (0.19-9.33) 4/17 (24) 2/17 (12) 1.18 (0.79-1.44) 1/4 (25) 2.41 (0.65-9.39) 12/17 (71) 10/17 (59) 6/17 (35) 6.19 (0.53-26.33) 1.48 (0.31-25.91) 12/17 (71) 5/17 (29) 2.48 (0.32-23.81) 3.26 (0.42-10.10) 1.41 (0.58-2.65) 1.66 (0.85-3.71) 1.44 (0.33-8.21) 10/16 (63) 1.21 (0.74-3.06)
(0.64; 1.74) (0.90; 1.25) (0.84; 1.30) (0.50; 0.34) (0.87; 1.28) (0.24; 2.22) (0.88; 1.12) (0.98; 1.25) (0.94; 1.28) (2.71; 4.71) (1.19; 5.69) (0.54; 4.82) (0.49; 4.56) (0.98; 5.07) (1.10; 2.28) ( 0.90; 0.12) (0.47; 0.35) (0.62; 1.65) (0.81; 1.08) (0.13; 0.80)
183 (99) 67/72 (93) 26/77 (34) 9/131 (7) 3/66 (4) 95 (52) 31 (17) 3 (2)
20 (100) 7/7 (100) 8/11 (73) 2/16 (13) 0/6 (0) 8 (40) 2 (10) 0 (0)
(1.06; 1.16) (1.03; 1.19) (1.28; 21.39) (0.38; 9.87) N.D. (0.24; 1.60) (0.12; 2.48) N.D.
N.S. N.S. 0.020 N.S. N.S. N.S. N.S. N.S.
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S.
80 (73)/30 (27) 83 (73)/30 (27) 88 (69)/40 (31)
3 (33)/6 (67) 6 (43)/8 (57) 3 (19)/13 (81)
(1.25; 22.69) (1.18; 11.51) (2.57; 35.33)
0.022 0.028 < 0.001
N.S. N.S. < 0.001
48 (26)
6 (30)
(0.43; 3.24)
N.S.
N.S.
PBC (n = 184) 52 (12-87) 165 (90) 6.8 (0.5-33.)
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.014 N.S. N.S. N.S. < 0.001
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. < 0.002
PBC/AIH patients showed lower serum levels of CHE, higher levels of IgG, advanced disease stages, and higher degrees of fibrosis (Pc > 0.050). After Bonferroni correction, only a higher degree of inflammation in PBC/AIH patients was significantly different (Pc < 0.001). All data are presented as median (minimum to maximum), if not otherwise indicated. *95% confidence interval of the difference (equal variances not assumed) or 95% confidence interval for cohort analyses. wP-value corrected for multiple testing after Bonferroni. ALP indicates alkaline phosphatase; AMA, anti-mitochondrial antibodies; ANA, anti-nuclear antibodies; AUROC, area under the ROC curve; CHE, cholinesterase; CI, confidence interval; GOT, glutamate oxaloacetate transaminase; GPT, glutamate pyruvate transaminase; Ig, immunoglobulin; LLN, lower limit of normal; N.D., not defined; N.S., not significant (P > 0.05); PBC, primary biliary cirrhosis; PBC/AIH, primary biliary cirrhosis/autoimmune hepatitis; SLA, soluble liver antigen; SMA, Smooth muscle antigen; ULN, Upper limit of normal.
period. A high specificity (92%) and a very high NPV (97%) underline the value of the MELD score in PBC patients. However, a moderate sensitivity (60%) and low PPV (35%) should be considered. By correlating the APRI score with the histologic findings, we detected significantly higher APRI values in patients with advanced fibrosis at the time of diagnosis (Pc = 0.022, Fig. 3). Patients with an APRI score >1.5 [10 of 126 patients (8%)] had a 12.9 times higher incidence of bridging fibrosis or cirrhosis, compared with patients with a value 6 in 13 of 80 patients (16%) and was associated with a 1.6-fold higher risk (95% CI, 1.06-2.50; Pc < 0.001) of LT or death, compared with patients with an Rvalue 0.050). PBC/AIH indicates primary biliary cirrhosis/autoimmune hepatitis.
liver-associated mortality or underwent LT during 14 years of observation (Fig. 5). Three of the 13 high-risk patients died and 2 required LT. Sensitivity and NPV of the Mayo model were 100% and the specificity was 89%. Despite a high R-value, only 5 of 13 patients showed complications, the PPV turned out to be low (38%). We calculated the expected probability of survival with the European model after 10 years. Interestingly, patients presented with a significantly worse clinical course compared with the outcome predicted by the European Model (Fig. 6). Sixty-four patients were evaluated and divided into 2 groups: a low-risk group for LT or death [probability of survival >80% after 10 years of therapy, 58 patients (91%)] and a high-risk group [probability of survival 0.050). MELD indicates Model for End-stage Liver Disease; PBC/AIH, primary biliary cirrhosis/autoimmune hepatitis.
442 | www.jcge.com
20%
40%
60%
80%
100%
1 - Specifity
FIGURE 4. Receiver Operating Characteristic curve of APRI score according to early (Desmet/Scheuer 0-2) or late stage fibrosis (Desmet/Scheuer 3-4). AUC is 0.77; 95% CI, 0.67-0.86; P < 0.001. APRI indicates Aspartate aminotransferase to Platelet Ratio Index; CI, confidence interval.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
Volume 49, Number 5, May/June 2015
Predictive Scores in PBC
100%
100%
90%
90%
Estimated survival [%]
Estimated survival [%]
J Clin Gastroenterol
80%
70%
Mayo risk score:
60%
0
2
4
6
8
10
12
Yale model low risk middle risk high risk low risk (censored) middle risk (censored) high risk (censored)
50% 0
14
2
4
6
8
10
12
14
16
18
20
Time [years]
Time [years]
FIGURE 5. Transplantation-free survival of patients with PBC or PBC/AIH according to their Mayo risk score (MRS).40 Patients with MRS < 6 showed a 10-year survival of 100%, whereas 5 of 13 patients (38%) with MRS > 6 received liver transplantation (2 patients) or died (3 patients) after 5 years (P < 0.001). PBC/AIH indicates primary biliary cirrhosis/autoimmune hepatitis.
group, 5 of 21 patients died (24%) and 3 patients received LT (14%). We calculated an NPV of 93% and a specificity of 80%. Sensitivity (53%) and PPV (27%) were low implying that only a fourth of the medium-risk or high-risk patients developed complications in the further course. No significant difference could be detected (Pc > 0.05). In addition, 58 patients were evaluated according to the criteria of Angulo et al.43 A decrease of ALP < 2 ULN after 6 months of UDCA therapy was observed in 50 patients (86%). The decrease of ALP was significant from 1.42 (0.32 to 5.11) ULN to 0.99 (0.46 to 2.00) ULN (P < 0.003). Indeed, all patients with ALP decrease 6 R-value < 6 censored R-value > 6 censored
50%
80%
FIGURE 7. Estimated survival of patients with low, medium, or high risk according to the Yale model at study entry.42 High-risk patients showed the most favorable outcome, whereas patients with medium risk had the worst prognosis (P < 0.001).
All 6 patients with PBC/AIH showed no decrease in ALP under UDCA monotherapy indicating poor clinical outcome. After 7 years, 1 patient died because of liverassociated complications, 1 patient required LT (33%). Thirty-five patients were evaluable for the Barcelona criteria. Serum ALP levels were compared before and after 1 year of UDCA treatment. Twenty-five of 35 patients (71%) fulfilled the criteria and showed a significant decrease of ALP (P < 0.001). During the observation period of up to 5 (range, 1.17 to 17.50) years, no patient died or received LT. A correlation with histologic staging was not possible because of the limited number of patients. The Corpechot score was determined in 39 evaluable patients and 35 patients (90%) fulfilled the criteria for a median observational period of 4.25 (range, 1.17 to 7.50) years. We observed more patients with a response to UDCA (90% compared with 61% according to Corpechot and colleagues). Patients with histologic stage III-IV according to Ludwig showed nearly identical rates of treatment failure (67% vs. 64%). A summary of all scores and models is presented in Table 2.
DISCUSSION 80%
70%
European model 60%
Patients with PBC, PBC/AIH Predicted outcome (European model) Patients with PBC, PBC/AIH (censored) Predicted outcome (European model, censored)
50% 0
1
2
3
4
5
6
7
8
9
10
Time [years]
FIGURE 6. Estimated and actual transplantation-free survival in patients with PBC or PBC/AIH. Survival rate according to the European model was significantly superior to the outcome of the cohort, P = 0.006.41 PBC/AIH indicates primary biliary cirrhosis/ autoimmune hepatitis.
Copyright
r
Despite increasing prevalence, PBC remains a rare chronic inflammatory autoimmune disease of unknown etiology primarily affecting interlobular bile ducts.50,51 An OS with AIH is described in 4.8% to 19% of patients with PBC.22–31 UDCA, a naturally occurring human bile acid, has been used for therapy for >2 decades under the assumption that it accelerates the intrahepatic transport and canalicular excretion of toxic hydrophobic bile acids in the liver. Patients with a response to UDCA treatment show a significantly improved survival44,52 but no cure is achieved.53 Early stages of PBC better respond to treatment than later stages. However, even in later stages, improved LT-free survival was described.54 Nowadays, UDCA is widely used in all disease stages, except in the presence of decompensated cirrhosis or pregnancy. Patients with PBC/AIH often show an incomplete response to UDCA treatment with a higher risk of developing decompensated liver cirrhosis.55
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.jcge.com |
443
J Clin Gastroenterol
Weinmann et al
Volume 49, Number 5, May/June 2015
TABLE 2. Evaluation of Non-PBC-specific Scores
Criteria Non-PBC-specific scores Child-Pugh A/B, C MELD 6-7/Z8 APRI < 1.0/ >1.5 PBC-specific models MRS R-value 0 Yale Low/middle, high risk PBC-specific scores for UDCA therapy Angulo ALP < 2ULN after 6 mo Pare´s Corpechot
Normalization or 40% decrease of ALP after 1 y ALP
Predictive Scores in Primary Biliary Cirrhosis A Retrospective Single Center Analysis of 204 Patients Arndt Weinmann, MD, MBA,*w Thomas Sattler, MD,z Hans-Peter Unold, MD,y Annette Grambihler, MD,w Andreas Teufel, MD, PhD,8 Sandra Koch,*w Marcus Schuchmann, MD, PhD,z Stefan Biesterfeld, MD, PhD,# Marcus A. Wo¨rns, MD, PhD,*w Peter R. Galle, MD, PhD,w and Henning Schulze-Bergkamen, MD, PhD**
Goals: The aim of this study was to assess the long-term outcome of primary biliary cirrhosis (PBC) patients and to test the clinical value of various outcome models, such as the Mayo Risk Score (MRS), in a large single-center cohort in Germany. Background: PBC is a chronic autoimmune liver disease with a female gender predominance and a peak incidence in the fifth decade of life. PBC is characterized by portal inflammation and immunemediated destruction of intrahepatic bile ducts in liver histology and the presence of antimitochondrial antibodies in the serum of nearly 95% of patients. In 5% to 20% of patients an overlap syndrome with autoimmune hepatitis (AIH) is diagnosed. Ursodeoxycholic acid is widely accepted as the standard medical treatment. Study: A total of 204 patients with PBC or PBC/AIH were retrospectively analyzed with regard to their clinical, biochemical, serological, and histologic features. PBC was diagnosed on the basis of the American Association for the Study of Liver Diseases criteria. Specific PBC scores, such as the MRS, the European and the Yale model, as well as nonspecific scores such as the ChildPugh, the Model for End-stage Liver Disease, and Aspartate Aminotransferase to Platelet Ratio Index score were analyzed for their utility to predict the clinical outcome of patients. Results: One hundred eighty-four patients with PBC alone and 20 with primary biliary cirrhosis/autoimmune hepatitis overlap were followed up for an average of 7.0 (range, 0.5 to 33.2) years. Importantly, baseline values of serum bilirubin, alkaline phosphatase, immunoglobulin M (IgM) and IgG, as well as antimitochondrial antibodies titers did not allow in properly predicting patient’s outcome. The MRS proved clinical applicability. Patients Received for publication November 14, 2013; accepted May 22, 2014. From the *Clinical Registry Unit (CRU); wDepartment of Internal Medicine I, University Medical Center of the Johannes Gutenberg University Mainz, Mainz; zDepartment of Surgery; **Department of Medical Oncology, National Center for Tumor Diseases (NCT), University Clinic of Heidelberg, Heidelberg; yDepartment of Surgery, Kreuznacher Diakonie Hospital, Bad Kreuznach; 8Department of Internal Medicine I, University Clinic of Regensburg, Regensburg; zDepartment of Internal Medicine, Hospital Konstanz, Konstanz; and #Department of Pathology, University Clinic of Du¨sseldorf, Du¨sseldorf, Germany. A.W. and T.S. contributed equally. A.W., T.S., H.-P.U., H.S.-B.: manuscript preparation and data and statistical analysis; T.S., H.-P.U., A.G., A.T., S.B., M.S., and M.A.W.: data collection; A.W., S.K., and M.A.W.: clinical registry database; A.T., M.S., S.B., M.A.W., and P.R.G.: revision of manuscript; and P.R.G.: study supervision. P.R.G. has received lecture fees from the Dr Falk Pharma GmbH. The remaining authors declare that they have nothing to disclose. Reprints: Arndt Weinmann, MD, MBA, Department of Internal Medicine I, Johannes Gutenberg-University Mainz, Langenbeckstrae 1, Mainz 55101, Germany (e-mail: [email protected]). Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved.
438 | www.jcge.com
with an R-value 40% from baseline after 1 year of UDCA treatment was associated with improved prognosis.21 An overlap syndrome (OS) with autoimmune hepatitis (AIH) has been described in 4.8% to 19% of PBC patients.22–31 Patients present with symptoms like fatigue, pruritus, jaundice, arthralgia, and myalgia. Primary biliary cirrhosis/autoimmune hepatitis (PBC/AIH), like PBC itself, is more common in women. Identification of patients with OS is clinically relevant as these patients have a poorer outcome and often respond to immunosuppressive therapy.23,25 In the absence of standardized diagnostic criteria, Chazouille`res and colleagues introduced a score for diagnosis of PBC/AIH.23 This retrospective single center study analyzed comprehensive clinical, serological, and histologic data of 204 patients with PBC and PBC/AIH undergoing UDCA treatment with a median observation period of 7 years. Various prognostic models and scores were applied to predict patient outcome and to test their clinical applicability.
MATERIALS AND METHODS Patient Population All consecutive patients with confirmed PBC with sufficient clinical data that were treated in the University Medical Center of the Johannes Gutenberg University Mainz between January 1990 and December 2009 were included in this retrospective analysis. The center is a quaternary referral center for autoimmune and other chronic liver diseases, treating more advanced stages of liver diseases and also conducting LT. Patients who fulfilled the criteria established by the American Association for the Study of Liver Diseases19 were included in this study. Study entry was defined as the date of the first presentation in our department. No formal ethics approval was required for this strictly retrospective study as was ruled by the local ethics committee (EthikKommission der Landesa¨rztekammer Mainz). Data were stored and analyzed in a specialized clinical registry database developed for this study. Access was strictly limited to the study investigators. Diagnosis of PBC required at least 2 of the 3 following criteria: (1) biochemical evidence of cholestasis [elevated levels of ALP and g-glutamyl transferase (gGT) over a period of 6 mo], (2) serological presence of AMA, and (3) a liver biopsy compatible with PBC or PBC/AIH. Measurement of AMA was performed using indirect immunofluorescence on rat tissue, whereas AMA subtyping was performed using enzyme-linked immunosorbent assay. An overlap-syndrome of PBC with AIH was diagnosed, if both criteria for PBC and AIH were fulfilled. AIH criteria included elevated immunoglobulin g (IgG) levels, serological presence of antinuclear antibodies (ANA), smooth muscle antibodies or soluble liver antigen, a liver biopsy compatible with AIH or PBC/AIH, as well as the exclusion of chronic viral hepatitis.32 Patients underwent regular physical examinations, laboratory assessments [eg, serum bilirubin, serum albumin, ALP, gGT, aspartate aminotransferase (AST), and alanine aminotransferase (ALT)], liver ultrasound, and upper Copyright
r
Predictive Scores in PBC
digestive endoscopy in cases of portal hypertension. Some patients received second liver biopsy to assess progression of liver disease. The histologic stage was assigned according to the Ludwig classification, and early-stage disease was defined as portal or periportal hepatitis in the absence of fibrotic septa or cirrhosis.33 Fibrosis and inflammation levels were classified according to Desmet et al34 and Scheuer.35 UDCA was prescribed with a daily dose of 13 to 15 mg/kg body weight, if advanced liver cirrhosis and pregnancy were excluded. Survival was analyzed as LT-free survival.
Prognostic Scores and Models Child-Pugh and the Model for End-stage Liver Disease (MELD) score were assessed as described.36–38 The Aspartate Aminotransferase to Platelet Ratio Index (APRI) was developed to assess the extent of liver fibrosis in viral hepatitis. It was calculated as following39: APRI = [glutamate oxaloacetate transaminase (GOT)] [upper limit of normal (ULN)] 100)/platelet count [109/L]. The Mayo Risk Score (MRS) was calculated as described40: R-value = 0.871 ln[serum bilirubin (mg/ dL)] 2.53 ln[albumin (mg/dL)] + 0.039 age + 6.843 + 1.02 ln(International Normalized Ratio) + 0.859 ascites (no = 0, yes = 1). The European model was measured as following41: 2.53 (log(serum bilirubin [mmol/dL]) 1.53) + 1.39 ascites (no = 0, yes = 1) 0.085 [albumin (g/L) 34.3] + 0.04 (age–55) + 0.65 upper gastrointestinal bleeding (no = 0, yes = 1). The Yale model was calculated42: 0.037 1.037(age) + 0.74 2.1 hepatomegaly (no = 0, yes = 1) + 0.82 2.26 [serum bilirubinZ5 mg/dL (no = 0, yes = 1)] 0.73 0.48 [serum bilirubin 40% or normalization of ALP after 1 year of UDCA therapy and to Corpechot et al45 (ALP < 3 ULN, GOT 1ULN [n(%)] > 2ULN [n(%)] > 3ULN [n(%)] GGT (ULN) GOT (ULN) > 1ULN [n(%)] > 2ULN [n(%)] GPT (ULN) GLDH (ULN) CHE (LLN) Platelets (LLN) IgM (ULN) > 1ULN [n(%)] IgG (ULN) Serological findings AMA positive [n(%)] AMA/M2 positive [n(%)] ANA positive [n(%)] SMA-positive [n(%)] SLA positive [n(%)] Anti-HAV positive [n(%)] HBsAg positive [n(%)] Anti-HCV positive [n(%)] Histologic findings Ludwig score (I-II/III-IV) Fibrosis (0-2/3-4) Inflammation (0-2/3-4) Other features Splenomegaly [n(%)]
PBC/AIH (n = 20)
95% CI*
P
Pcw
55 (26-71) 18 (90) 8.5 (2.3-32.0)
( 7; 5) (0.21; 4.48) ( 1.0; 5.8)
N.S. N.S. N.S.
N.S. N.S. N.S.
0.50 (0.15-12.08) 23/142 (16) 12/142 (8) 1.24 (0.79-1.56) 10/78 (13) 1.46 (0.28-7.28) 104/145 (72) 52/145 (36) 30/145 (21) 4.79 (0.14-36.11) 1.23 (0.25-23.78) 85/142 (60) 31/142 (22) 1.42 (0.19-25.29) 2.47 (0.24-13.84) 2.02 (0.02-3.95) 1.73 (0.25-4.07) 1.42 (0.27-9.79) 102/136 (75) 0.84 (0.37-2.05)
0.75 (0.19-9.33) 4/17 (24) 2/17 (12) 1.18 (0.79-1.44) 1/4 (25) 2.41 (0.65-9.39) 12/17 (71) 10/17 (59) 6/17 (35) 6.19 (0.53-26.33) 1.48 (0.31-25.91) 12/17 (71) 5/17 (29) 2.48 (0.32-23.81) 3.26 (0.42-10.10) 1.41 (0.58-2.65) 1.66 (0.85-3.71) 1.44 (0.33-8.21) 10/16 (63) 1.21 (0.74-3.06)
(0.64; 1.74) (0.90; 1.25) (0.84; 1.30) (0.50; 0.34) (0.87; 1.28) (0.24; 2.22) (0.88; 1.12) (0.98; 1.25) (0.94; 1.28) (2.71; 4.71) (1.19; 5.69) (0.54; 4.82) (0.49; 4.56) (0.98; 5.07) (1.10; 2.28) ( 0.90; 0.12) (0.47; 0.35) (0.62; 1.65) (0.81; 1.08) (0.13; 0.80)
183 (99) 67/72 (93) 26/77 (34) 9/131 (7) 3/66 (4) 95 (52) 31 (17) 3 (2)
20 (100) 7/7 (100) 8/11 (73) 2/16 (13) 0/6 (0) 8 (40) 2 (10) 0 (0)
(1.06; 1.16) (1.03; 1.19) (1.28; 21.39) (0.38; 9.87) N.D. (0.24; 1.60) (0.12; 2.48) N.D.
N.S. N.S. 0.020 N.S. N.S. N.S. N.S. N.S.
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S.
80 (73)/30 (27) 83 (73)/30 (27) 88 (69)/40 (31)
3 (33)/6 (67) 6 (43)/8 (57) 3 (19)/13 (81)
(1.25; 22.69) (1.18; 11.51) (2.57; 35.33)
0.022 0.028 < 0.001
N.S. N.S. < 0.001
48 (26)
6 (30)
(0.43; 3.24)
N.S.
N.S.
PBC (n = 184) 52 (12-87) 165 (90) 6.8 (0.5-33.)
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. 0.014 N.S. N.S. N.S. < 0.001
N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. N.S. < 0.002
PBC/AIH patients showed lower serum levels of CHE, higher levels of IgG, advanced disease stages, and higher degrees of fibrosis (Pc > 0.050). After Bonferroni correction, only a higher degree of inflammation in PBC/AIH patients was significantly different (Pc < 0.001). All data are presented as median (minimum to maximum), if not otherwise indicated. *95% confidence interval of the difference (equal variances not assumed) or 95% confidence interval for cohort analyses. wP-value corrected for multiple testing after Bonferroni. ALP indicates alkaline phosphatase; AMA, anti-mitochondrial antibodies; ANA, anti-nuclear antibodies; AUROC, area under the ROC curve; CHE, cholinesterase; CI, confidence interval; GOT, glutamate oxaloacetate transaminase; GPT, glutamate pyruvate transaminase; Ig, immunoglobulin; LLN, lower limit of normal; N.D., not defined; N.S., not significant (P > 0.05); PBC, primary biliary cirrhosis; PBC/AIH, primary biliary cirrhosis/autoimmune hepatitis; SLA, soluble liver antigen; SMA, Smooth muscle antigen; ULN, Upper limit of normal.
period. A high specificity (92%) and a very high NPV (97%) underline the value of the MELD score in PBC patients. However, a moderate sensitivity (60%) and low PPV (35%) should be considered. By correlating the APRI score with the histologic findings, we detected significantly higher APRI values in patients with advanced fibrosis at the time of diagnosis (Pc = 0.022, Fig. 3). Patients with an APRI score >1.5 [10 of 126 patients (8%)] had a 12.9 times higher incidence of bridging fibrosis or cirrhosis, compared with patients with a value 6 in 13 of 80 patients (16%) and was associated with a 1.6-fold higher risk (95% CI, 1.06-2.50; Pc < 0.001) of LT or death, compared with patients with an Rvalue 0.050). PBC/AIH indicates primary biliary cirrhosis/autoimmune hepatitis.
liver-associated mortality or underwent LT during 14 years of observation (Fig. 5). Three of the 13 high-risk patients died and 2 required LT. Sensitivity and NPV of the Mayo model were 100% and the specificity was 89%. Despite a high R-value, only 5 of 13 patients showed complications, the PPV turned out to be low (38%). We calculated the expected probability of survival with the European model after 10 years. Interestingly, patients presented with a significantly worse clinical course compared with the outcome predicted by the European Model (Fig. 6). Sixty-four patients were evaluated and divided into 2 groups: a low-risk group for LT or death [probability of survival >80% after 10 years of therapy, 58 patients (91%)] and a high-risk group [probability of survival 0.050). MELD indicates Model for End-stage Liver Disease; PBC/AIH, primary biliary cirrhosis/autoimmune hepatitis.
442 | www.jcge.com
20%
40%
60%
80%
100%
1 - Specifity
FIGURE 4. Receiver Operating Characteristic curve of APRI score according to early (Desmet/Scheuer 0-2) or late stage fibrosis (Desmet/Scheuer 3-4). AUC is 0.77; 95% CI, 0.67-0.86; P < 0.001. APRI indicates Aspartate aminotransferase to Platelet Ratio Index; CI, confidence interval.
Copyright
r
2014 Wolters Kluwer Health, Inc. All rights reserved.
Volume 49, Number 5, May/June 2015
Predictive Scores in PBC
100%
100%
90%
90%
Estimated survival [%]
Estimated survival [%]
J Clin Gastroenterol
80%
70%
Mayo risk score:
60%
0
2
4
6
8
10
12
Yale model low risk middle risk high risk low risk (censored) middle risk (censored) high risk (censored)
50% 0
14
2
4
6
8
10
12
14
16
18
20
Time [years]
Time [years]
FIGURE 5. Transplantation-free survival of patients with PBC or PBC/AIH according to their Mayo risk score (MRS).40 Patients with MRS < 6 showed a 10-year survival of 100%, whereas 5 of 13 patients (38%) with MRS > 6 received liver transplantation (2 patients) or died (3 patients) after 5 years (P < 0.001). PBC/AIH indicates primary biliary cirrhosis/autoimmune hepatitis.
group, 5 of 21 patients died (24%) and 3 patients received LT (14%). We calculated an NPV of 93% and a specificity of 80%. Sensitivity (53%) and PPV (27%) were low implying that only a fourth of the medium-risk or high-risk patients developed complications in the further course. No significant difference could be detected (Pc > 0.05). In addition, 58 patients were evaluated according to the criteria of Angulo et al.43 A decrease of ALP < 2 ULN after 6 months of UDCA therapy was observed in 50 patients (86%). The decrease of ALP was significant from 1.42 (0.32 to 5.11) ULN to 0.99 (0.46 to 2.00) ULN (P < 0.003). Indeed, all patients with ALP decrease 6 R-value < 6 censored R-value > 6 censored
50%
80%
FIGURE 7. Estimated survival of patients with low, medium, or high risk according to the Yale model at study entry.42 High-risk patients showed the most favorable outcome, whereas patients with medium risk had the worst prognosis (P < 0.001).
All 6 patients with PBC/AIH showed no decrease in ALP under UDCA monotherapy indicating poor clinical outcome. After 7 years, 1 patient died because of liverassociated complications, 1 patient required LT (33%). Thirty-five patients were evaluable for the Barcelona criteria. Serum ALP levels were compared before and after 1 year of UDCA treatment. Twenty-five of 35 patients (71%) fulfilled the criteria and showed a significant decrease of ALP (P < 0.001). During the observation period of up to 5 (range, 1.17 to 17.50) years, no patient died or received LT. A correlation with histologic staging was not possible because of the limited number of patients. The Corpechot score was determined in 39 evaluable patients and 35 patients (90%) fulfilled the criteria for a median observational period of 4.25 (range, 1.17 to 7.50) years. We observed more patients with a response to UDCA (90% compared with 61% according to Corpechot and colleagues). Patients with histologic stage III-IV according to Ludwig showed nearly identical rates of treatment failure (67% vs. 64%). A summary of all scores and models is presented in Table 2.
DISCUSSION 80%
70%
European model 60%
Patients with PBC, PBC/AIH Predicted outcome (European model) Patients with PBC, PBC/AIH (censored) Predicted outcome (European model, censored)
50% 0
1
2
3
4
5
6
7
8
9
10
Time [years]
FIGURE 6. Estimated and actual transplantation-free survival in patients with PBC or PBC/AIH. Survival rate according to the European model was significantly superior to the outcome of the cohort, P = 0.006.41 PBC/AIH indicates primary biliary cirrhosis/ autoimmune hepatitis.
Copyright
r
Despite increasing prevalence, PBC remains a rare chronic inflammatory autoimmune disease of unknown etiology primarily affecting interlobular bile ducts.50,51 An OS with AIH is described in 4.8% to 19% of patients with PBC.22–31 UDCA, a naturally occurring human bile acid, has been used for therapy for >2 decades under the assumption that it accelerates the intrahepatic transport and canalicular excretion of toxic hydrophobic bile acids in the liver. Patients with a response to UDCA treatment show a significantly improved survival44,52 but no cure is achieved.53 Early stages of PBC better respond to treatment than later stages. However, even in later stages, improved LT-free survival was described.54 Nowadays, UDCA is widely used in all disease stages, except in the presence of decompensated cirrhosis or pregnancy. Patients with PBC/AIH often show an incomplete response to UDCA treatment with a higher risk of developing decompensated liver cirrhosis.55
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.jcge.com |
443
J Clin Gastroenterol
Weinmann et al
Volume 49, Number 5, May/June 2015
TABLE 2. Evaluation of Non-PBC-specific Scores
Criteria Non-PBC-specific scores Child-Pugh A/B, C MELD 6-7/Z8 APRI < 1.0/ >1.5 PBC-specific models MRS R-value 0 Yale Low/middle, high risk PBC-specific scores for UDCA therapy Angulo ALP < 2ULN after 6 mo Pare´s Corpechot
Normalization or 40% decrease of ALP after 1 y ALP
