The Evolving Use of Prognostic Scores and the Role of Biochemical Markers in PBC
The role of biochemical markers in the management of PBC has evolved a great deal over the past several years. A variety of prognostic models have been relied on to optimize management of patients with PBC.1 The Mayo score had been considered the classic prognostic model for patients with untreated disease, and it could be used to describe the natural history of PBC.2 The Mayo score incorporates factors such as the patient’s age; their levels of bilirubin, aspartate transaminase (AST), and albumin; and the presence of variceal bleeding. Previously, this score and others were useful for the patients presenting for treatment in the clinic, who had later stages of disease. The increasing ability to diagnose PBC earlier in the disease course has made these prognostic models less relevant. The older models are most applicable to patients with more advanced disease.
As more patients are diagnosed earlier in the course of the disease—when their liver function is less compromised— there has been an effort to utilize biochemical markers as potential predictors of outcome in PBC.
The earlier diagnosis of PBC led to the realization that there is a difference between disease stage and prognostic risk category. This difference is highlighted by the use of the Rotterdam criteria, which combine bilirubin and albumin measurements to categorize patients into groups with different prognostic and survival outcomes.3 In this way, the Rotterdam criteria are focused on disease stage and the associated liver function.
Primary Biochemical Markers
As more patients are diagnosed earlier in the course of the disease—when their liver function is less compromised— there has been an effort to utilize biochemical markers as potential predictors of outcome in PBC. Thus far, the primary biochemical markers used in patients with PBC include bilirubin and ALP. Between these, it has become evident that bilirubin is a signifier of more advanced disease stages, whereas ALP might better predict long-term outcome and the risk for future events.
Serum bilirubin is well defined as an independent predictor of prognosis and the natural course of PBC, and this marker is incorporated into the Mayo PBC score.4-7 Despite the established prognostic utility of bilirubin, its application is limited to patients with relatively advanced disease, who are most likely to show meaningful changes in bilirubin levels. In contrast, the isoenzyme ALP appears to be more broadly applicable across the spectrum of PBC disease severity.8,9 Elevated levels of ALP are a marker of cholestasis.
The Global PBC Study
The correlation of serum ALP and bilirubin levels—either individually or in combination—with transplant-free survival was evaluated by Lammers and colleagues in the Global PBC Study.10 This large, international, observational PBC database was powered to permit an individual patient-level meta-analysis to determine the prognostic significance of these biochemical markers. Data from the Global PBC Study Group collaboration represented 15 liver centers in 8 North American and European countries, each of which contributed sets of patient data from major long-term follow-up cohorts. The majority of follow-up data were collected from patients initiating UDCA therapy.
A total of 4845 patients were included in the analysis.10 Patients had been diagnosed with PBC between 1959 and 2012; 79% had received their diagnosis after 1990. Patients were followed for a median of 7.3 years (interquartile range [IQR], 3.6-11.5). The histologic disease stage was reported for the patients who had undergone a liver biopsy (76%); most of these patients had stage I or II disease. According to the Rotterdam criteria, the biochemical disease stage was early in 42%, moderately advanced in 15%, and advanced in 5% (the stage was not available in 38%). At baseline, the median serum ALP level was 2.10 (IQR, 1.31-3.72; the level was not available in 24%). The median serum bilirubin level was 0.67 (IQR, 0.45-1.06; the level was not available in 23%).
In the total cohort, the transplant-free survival rate was 88% at 5 years, 77% at 10 years, and 63% at 15 years.10 A total of 85% of patients were treated with UDCA. Among these patients, the rates of transplant-free survival were 90% at 5 years, 78% at 10 years, and 66% at 15 years. These rates were significantly higher than those reported in untreated patients (79%, 59%, and 32%, respectively; P<.0001).
The levels of both ALP and bilirubin at baseline and each year over 5 years were associated with the risk for liver transplant and death (Figures 3 and 4).10 Higher levels were associated with worse clinical outcomes. A threshold of 2.0 × the upper limit of normal (ULN) for serum ALP was found to predict clinical outcomes. For patients with ALP levels at or less than 2.0 × ULN, the rates of transplant-free survival were 94% at 5 years, 84% at 10 years, and 73% at 15 years. In comparison, for patients with ALP levels higher than 2.0 × ULN, these rates were 81%, 62%, and 50%, respectively (P<.0001). Similarly, the 5-year, 10-year, and 15-year transplant-free survival rates were higher in patients with normal bilirubin levels after 1 year of follow-up (95%, 86%, and 74%, respectively) vs those with abnormal bilirubin levels (65%, 41%, and 30%, respectively). These differences in transplant-free survival associated with the bilirubin level were also statistically significant (P<.0001). The association between elevated ALP levels and worse transplant-free survival was significantly higher in patients with both normal and abnormal bilirubin levels.10
Figure 3.
The hazard for liver transplant or death according to alkaline phosphatase levels at different time points estimated with cubic spline function in the Global PBC Study. ULN, upper limit of normal. aAmong 4635 patients, 3710 were included in this analysis. bAmong 3161 patients, 2203 were included in this analysis. Adapted from Lammers WJ et al. Gastroenterology. 2014;147(6):1338-1349.e5.10
Figure 4.
The hazard for liver transplant or death according to bilirubin levels at different time points estimated with cubic spline function in the Global PBC Study. aAmong 4635 patients, 3681 were included in this analysis. bAmong 3161 patients, 2109 were included in this analysis. Adapted from Lammers WJ et al. Gastroenterology. 2014;147(6):1338-1349.e510
The Global PBC database meta-analysis demonstrated a strong association between abnormally increased serum ALP and bilirubin levels with reduced transplant-free survival in patients with PBC. Furthermore, this analysis confirmed that a combination of both variables improves prognostic prediction in these patients, regardless of whether they received UDCA.10
Evidence for Prognostic Scores to Determine Response to First-Line UDCA Therapy
UDCA is a frequent first-line treatment in patients with PBC, and it improves survival. The biochemical response to treatment with UDCA, referred to as the treatment response, is a strong predictor of long-term outcomes in patients with PBC.11 This knowledge led to the development of several prognostic models based on biochemical response. Several of these prognostic scores have been validated as highly accurate in this setting, and they are widely used to risk-stratify patients with PBC. There are now a number of different response criteria prognostic models that are based on the concept of how the patient is doing after 1 or 2 years of UDCA therapy. These models, which initially used varying combinations of bilirubin and ALP levels to determine response, have since incorporated other biochemical markers to predict long-term outcomes with regard to the risk for liver transplant or death.
GLOBE PBC Score
The Global PBC Study Group aimed to identify UDCA-treated patients with an insufficient response to treatment.12 The investigators compared a dataset of PBC patients against a representative healthy population to develop the GLOBE PBC score. The score was developed based on data from the patients treated with UDCA. Among these 2488 patients, the 5-year, 10-year, and 15-year transplant-free survival rates were 90.0%, 77.5%, and 65.6% respectively.12
Using this patient dataset, the GLOBE score incorporated age, bilirubin, albumin, ALP, and platelet count as independent predictors of liver transplant or death. This score was then applied to a validation cohort of 1631 patients who had overall characteristics and transplant-free survival rates that were similar to the derivation cohort.12
Patients with a GLOBE score above 0.30 (considered nonresponders; approximately 40% of patients) had a significantly diminished survival compared with a matched general population (hazard ratio [HR], 5.51; 95% CI, 4.52-6.72; P<.0001).12 The 5-year, 10-year, and 15-year transplant-free survival rates in this group of patients were 79.7%, 57.4%, and 42.5%, respectively. In comparison, patients with a GLOBE score of 0.30 or less (responders) had a life-expectancy comparable with that of a matched general population. The 5-year, 10-year, and 15-year transplant-free survival rates were 98.0%, 92.0%, and 82.3%, respectively (P<.0001). Nonresponsive patients were also significantly more likely to present with a late stage of disease at baseline than responding patients. The investigators concluded that the GLOBE score could reliably determine the prognosis of patients with PBC who have been treated with UDCA for 1 year, regardless of their disease stage.12
The UK-PBC Risk Score
The UK-PBC risk score was designed to estimate the absolute risk for developing end-stage liver disease among patients with PBC treated with UDCA.13 To develop the model, 1916 patients treated with UDCA were selected from the UK-PBC Research Cohort. At baseline, approximately 10% of these patients had advanced disease at diagnosis (defined by splenomegaly or ascites), and approximately 20% of participants were antinuclear antibody–positive. The UK-PBC risk score was developed using this derivation cohort to include 5 variables: albumin level, platelet level, level of bilirubin after 12 months of UDCA, levels of transaminases after 12 months of treatment, and level of ALP after 12 months of treatment.13
The model was applied to a validation cohort of 1249 patients treated with UDCA.13 Within the validation cohort, the area under the receiver operating characteristic curves (AUROCs) were 0.96 (95% CI, 0.93-0.99) for the 5-year risk score, 0.95 (95% CI, 0.93-0.98) for the 10-year risk score, and 0.94 (95% CI, 0.91-0.97) for the 15-year risk score (Figure 5).
Figure 5.
The predicted vs observed risk for an event across each decile of the UK‐PBC risk score at 5 years (A), 10 years (B) and 15 years (C). Adapted from Carbone M et al. Hepatology. 2016;63(3):930-950.13
The authors from this study concluded that in clinical practice, the UK-PBC scoring system could be useful to identify those patients at highest risk for developing end-stage liver disease and thus who would obtain the greatest benefit from further risk reduction using second-line therapy.13
The Paris-I/II Criteria
It is widely considered that serum ALP and bilirubin are the 2 most important parameters in evaluating response to UDCA.14 The Paris-I criteria were developed to discriminate between low- and high-risk patients treated with UDCA. The Paris-I criteria are generally considered to be a strong predictor of transplant-free survival in patients with PBC, and have been validated in large studies.15 The Paris-I criteria were defined as serum bilirubin at or less than 1 mg/dL, ALP at or below 3 ULN, and AST at or below 2 ULN, all assessed at 1 year after initiation of UDCA therapy. Death or liver transplant were determined to be 2.5-times more likely to occur in patients who showed either ALP exceeding 3 × ULN, AST exceeding × 2 ULN, or serum bilirubin higher than 1 mg/dL at 1 year of treatment. This patient subgroup—with a high risk for liver transplant or death—accounts for nearly 40% of all patients and has a 10-year transplant-free survival rate of approximately 50%. In contrast, survival rates among patients without these elevated biochemical markers were similar to those of a control population.15
Additionally, as disease stage is known to affect the biochemical response to UDCA, stage-specific thresholds were incorporated into the Paris-II criteria to better fit early-stage patients, an increasingly large proportion of patients with PBC. It was shown that patients meeting the Paris-II criteria, defined as patients with both ALP and AST at or less than 1.5 × ULN and normal total bilirubin after 1 year of UDCA therapy, had no evidence of progressive disease over an average of 7 years. Using the Paris-II criteria, adverse outcomes were observed only in nonresponding patients. The survival rates without adverse outcome at 5, 10, and 15 years of follow-up were 100% in responders. In nonresponders, these survival rates were 93%, 87%, and 74%, respectively.9
The Paris-II criteria were also evaluated to determine if they could be applicable to a population of patients with disease at a late histologic stage.9 However, the only criteria that were able to significantly discriminate among these patients with advanced PBC were the Paris-I criteria (HR in nonresponders, 1.39; 95% CI, 1.05-1.84; P<.05). In contrast, the Paris-II criteria did not reach statistical significance. This observation led the study authors to conclude that the Paris-II criteria should be limited to patients with early stages of the disease. Although the Paris-I criteria seemed more broadly applicable across PBC disease stages, they were found to be less accurate and reliable than the Paris-II criteria for early stages.9
Toronto Criteria
The Toronto criteria can predict histologic progression in patients with PBC.16 Using 10-year histologic progression of disease in paired biopsies from the same patient, combined with biochemical response to UDCA and baseline histology, the Toronto criteria were developed to predict the disease course in a cohort of patients with predominantly early disease. The Toronto criteria define biochemical response to UDCA as ALP less than 184 IU/L (1.67 × ULN) after 2 years of treatment. In paired liver biopsies, more than 80% of patients who did not respond to UDCA according to the Toronto criteria showed histologic progression after 10 years (odds ratio, 12.14; 95% CI, 2.69-54.74).16 Since its development in 2010, the Toronto criteria have become a platform for the design and conduct of clinical trials for the second-line treatment of PBC.
APRI
The AST to platelet ratio index (APRI) also predicts outcomes in PBC, and was shown to be independent of UDCA response.17 The clinical utility of APRI was suggested by a derivation cohort of 386 patients with PBC (AUROC, 0.781; 95% CI, 0.721-0.840). Here, an APRI higher than 0.54 at baseline was predictive of liver transplant or death (adjusted HR, 2.40; 95% CI, 1.32-4.36; P<.01). Importantly, the APRI continued to be statistically significantly predictive of liver transplant or death when applied at 1 year (adjusted HR, 2.75; 95% CI, 1.49-5.08; P<.01), despite controlling for UDCA-response (AUROC, 0.806; 95% CI, 0.756-0.857).17
A higher APRI at baseline (>0.54) was confirmed to predict adverse outcome (liver transplant or death) in the validation series, with sensitivity and specificity comparable with that observed in the derivation cohort (adjusted HR, 3.04; 95% CI, 1.66-5.55; P<.001). In the validation cohort of 629 patients, the AUROC for the APRI was 0.741 (95% CI, 0.662-0.820). The association of higher (>0.54) APRI with liver transplant or death in the validation cohort at 1 year was also significant (adjusted HR, 4.66; 95% CI, 2.39-9.12; P<.00001). At 1 year, the AUROC for the APRI in the validation cohort was 0.783 (95% CI, 0.709-0.857).17
By using a combination of APRI at 1 year and UDCA-response criteria, the investigators were able to classify patients as low, intermediate, or high risk. Patients with PBC defined as low risk had a biochemical response and an APRI at 1 year of 0.54 or less. Those defined as intermediate risk had either a biochemical response and an APRI at 1 year exceeding 0.54 or a biochemical nonresponse and an APRI at 1 year of 0.54 or less. Patients defined as high risk had a biochemical nonresponse and an APRI at 1 year of greater than 0.54.17
Transplant-free survival was significantly longer in low-risk patients compared with intermediate-risk patients. Additionally, high-risk patients had the poorest transplant-free survival. Ten-year transplant-free survival rates across the low-, intermediate-, and high-risk groups from the derivation cohort were 86%, 59%, and 13%, respectively. This correlation was confirmed in the validation series.17 The authors of this APRI model concluded that APRI at diagnosis and/or after 1 year could be used to independently predict liver transplant or death in patients with PBC.
New Data From the Global PBC Study Group
There have been 2 other more recent observations from the Global PBC Study Group. First is the recognition that a bilirubin threshold of 0.6 mg/dL may be a cutoff point at which the risk for liver transplant or death begins to increase. A recent publication from the Global PBC Study Group reported that a bilirubin threshold of 0.6 × ULN had the highest ability to predict liver transplant or death at 1 year (HR, 2.12; 95% CI, 1.69-2.66; P<.001).18 The 10-year survival rates of patients with a bilirubin level at or below 0.6 × ULN was 91.3%, whereas it was significantly lower for patients with a bilirubin level above the threshold of 0.6 × ULN (79.2%; P<.001). Furthermore, UDCA-induced reduction in bilirubin below this threshold was associated with an 11% improvement in 10-year survival.18
The second observation is that there may not be a dichotomous relationship between ALP and patient outcomes. Any ALP elevation exceeding 1 × ULN may indicate that the patient has an ascending linear risk with regard to long-term outcomes. Indeed, normalization of ALP levels appears to be the optimal goal, with 10-year survival rates of 93.2% in patients with ALP at or less than 1 × ULN and 86.1% in those with ALP between 1.0 to 1.67 × ULN.18
Physicians are starting to approach the treatment of PBC in a similar manner to that of autoimmune hepatitis. For example, the goal of treatment should be to achieve a complete biochemical remission, if possible. Of course, this goal has not been a possibility during treatment with FXR agonists because ALP normalization is not commonly observed with this mechanism of action. However, as the patient progresses to treatment with second-line and third-line therapies, as well as combination therapies, it is no longer acceptable to settle for an ALP of less than 1.67 × ULN, particularly in light of these latest data from the Global PBC Study Group. Instead, it may be better to strive for an ALP level as close to normal as possible, or at least below 1.5 × ULN.
Biochemical Markers Beyond ALP and Bilirubin
Additional data from the Global PBC Study Group suggest that GGT, a serum marker of cholestasis, may provide a further level of granularity.19 In a group of 2129 patients with PBC, there was a correlation between serum levels of GGT and ALP. Higher serum GGT levels were associated with a lower hazard for transplant-free survival. A threshold of GGT higher than 3.2 × ULN at 12 months after treatment identified those patients who required liver transplant or died from a liver-related cause at 10 years (AUROC, 0.70). The association between the risk for liver transplant or liver-related death and elevated serum GGT persisted even in patients with ALP levels below 1.5 × ULN. These data suggest that a serum GGT below 3.2 × ULN may be associated with additional prognostic value compared with ALP alone. Inclusion of information regarding the level of GGT increased the prognostic value of the GLOBE score.19
Liver stiffness, as assessed by transient elastography and magnetic resonance elastography, has also been explored for its prognostic ability. In a group of 538 patients with PBC, liver stiffness measurements by transient elastography (n=286) or magnetic resonance elastography (n=332) were reviewed.20 Liver stiffness cutoffs for predicting fibrosis stages were then determined using AUROC among those patients who underwent a liver biopsy. The optimal thresholds of liver stiffness for predicting advanced histologic stage (F4) were 14.40 kPa for transient elastography and 4.60 kPa for magnetic resonance elastography. Intriguingly, both measurements of liver stiffness were better able to predict histologic advanced fibrosis compared with biochemical markers. The ability of liver stiffness to predict hepatic decompensation remained even after adjustment for UDCA responsiveness (HR, 1.14; 95% CI, 1.05-1.24 for transient elastography and HR, 1.68; 95% CI, 1.28-2.19 for magnetic resonance elastography). Liver stiffness was also able to predict hepatic decompensation after adjusting for the GLOBE score (HR, 1.13; 95% CI, 1.07-1.19 for transient elastography and HR, 2.09; 95% CI, 1.57-2.78 for magnetic resonance elastography).20
Acknowledgment
This article was written by a medical writer based on a clinical roundtable discussion.
References
- 1.Chen S, Duan W, You H, Jia J. A brief review on prognostic models of primary biliary cholangitis. Hepatol Int. 2017;11(5):412–418. doi: 10.1007/s12072-017-9819-9. [DOI] [PubMed] [Google Scholar]
- 2.Kim WR, Therneau TM, Wiesner RH et al. A revised natural history model for primary sclerosing cholangitis. Mayo Clin Proc. 2000;75(7):688–694. doi: 10.4065/75.7.688. [DOI] [PubMed] [Google Scholar]
- 3.Kuiper EM, Hansen BE, de Vries RA et al. Dutch PBC Study Group. Improved prognosis of patients with primary biliary cirrhosis that have a biochemical response to ursodeoxycholic acid. Gastroenterology. 2009;136(4):1281–1287. doi: 10.1053/j.gastro.2009.01.003. [DOI] [PubMed] [Google Scholar]
- 4.Shapiro JM, Smith H, Schaffner F. Serum bilirubin: a prognostic factor in primary biliary cirrhosis. Gut. 1979;20(2):137–140. doi: 10.1136/gut.20.2.137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bonnand AM, Heathcote EJ, Lindor KD, Poupon RE. Clinical significance of serum bilirubin levels under ursodeoxycholic acid therapy in patients with primary biliary cirrhosis. Hepatology. 1999;29(1):39–43. doi: 10.1002/hep.510290140. [DOI] [PubMed] [Google Scholar]
- 6.Krzeski P, Zych W, Kraszewska E, Milewski B, Butruk E, Habior A. Is serum bilirubin concentration the only valid prognostic marker in primary biliary cirrhosis? Hepatology. 1999;30(4):865–869. doi: 10.1002/hep.510300415. [DOI] [PubMed] [Google Scholar]
- 7.Kim WR, Therneau TM, Wiesner RH et al. A revised natural history model for primary sclerosing cholangitis. Mayo Clin Proc. 2000;75(7):688–694. doi: 10.4065/75.7.688. [DOI] [PubMed] [Google Scholar]
- 8.Momah N, Silveira MG, Jorgensen R, Sinakos E, Lindor KD. Optimizing biochemical markers as endpoints for clinical trials in primary biliary cirrhosis. Liver Int. 2012;32(5):790–795. doi: 10.1111/j.1478-3231.2011.02678.x. [DOI] [PubMed] [Google Scholar]
- 9.Corpechot C, Chazouillères O, Poupon R. Early primary biliary cirrhosis: biochemical response to treatment and prediction of long-term outcome. J Hepatol. 2011;55(6):1361–1367. doi: 10.1016/j.jhep.2011.02.031. [DOI] [PubMed] [Google Scholar]
- 10.Lammers WJ, van Buuren HR, Hirschfield GM et al. Global PBC Study Group. Levels of alkaline phosphatase and bilirubin are surrogate end points of outcomes of patients with primary biliary cirrhosis: an international follow-up study. Gastroenterology. 2014;147(6):1338–1349.e5. doi: 10.1053/j.gastro.2014.08.029. [DOI] [PubMed] [Google Scholar]
- 11.Parés A, Caballería L, Rodés J. Excellent long-term survival in patients with primary biliary cirrhosis and biochemical response to ursodeoxycholic acid. Gastroenterology. 2006;130(3):715–720. doi: 10.1053/j.gastro.2005.12.029. [DOI] [PubMed] [Google Scholar]
- 12.Lammers WJ, Hirschfield GM, Corpechot C et al. Global PBC Study Group. Development and validation of a scoring system to predict outcomes of patients with primary biliary cirrhosis receiving ursodeoxycholic acid therapy. Gastroenterology. 2015;149(7):1804–1812.e4. doi: 10.1053/j.gastro.2015.07.061. [DOI] [PubMed] [Google Scholar]
- 13.Carbone M, Sharp SJ, Flack S et al. UK-PBC Consortium. The UK-PBC risk scores: derivation and validation of a scoring system for long-term prediction of end-stage liver disease in primary biliary cholangitis. Hepatology. 2016;63(3):930–950. doi: 10.1002/hep.28017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.European Association for the Study of the Liver. EASL Clinical Practice Guidelines: the diagnosis and management of patients with primary biliary cholangitis. J Hepatol. 2017;67(1):145–172. doi: 10.1016/j.jhep.2017.03.022. [DOI] [PubMed] [Google Scholar]
- 15.Corpechot C, Abenavoli L, Rabahi N et al. Biochemical response to ursodeoxycholic acid and long-term prognosis in primary biliary cirrhosis. Hepatology. 2008;48(3):871–877. doi: 10.1002/hep.22428. [DOI] [PubMed] [Google Scholar]
- 16.Kumagi T, Guindi M, Fischer SE et al. Baseline ductopenia and treatment response predict long-term histological progression in primary biliary cirrhosis. Am J Gastroenterol. 2010;105(10):2186–2194. doi: 10.1038/ajg.2010.216. [DOI] [PubMed] [Google Scholar]
- 17.Trivedi PJ, Bruns T, Cheung A et al. Optimising risk stratification in primary biliary cirrhosis: AST/platelet ratio index predicts outcome independent of ursodeoxycholic acid response. J Hepatol. 2014;60(6):1249–1258. doi: 10.1016/j.jhep.2014.01.029. [DOI] [PubMed] [Google Scholar]
- 18.Murillo Perez CF, Harms MH, Lindor KD et al. GLOBAL PBC Study Group. Goals of treatment for improved survival in primary biliary cholangitis: treatment target should be bilirubin within the normal range and normalization of alkaline phosphatase. Am J Gastroenterol. 2020;115(7):1066–1074. doi: 10.14309/ajg.0000000000000557. [DOI] [PubMed] [Google Scholar]
- 19.Gerussi A, Bernasconi DP, O’Donnell SE et al. Italian PBC Study Group and the GLOBAL PBC Study Group. Measurement of gamma glutamyl transferase to determine risk of liver transplantation or death in patients with primary biliary cholangitis. Clin Gastroenterol Hepatol. 2020;S1542-3565(20):31083–1. doi: 10.1016/j.cgh.2020.08.006. [DOI] [PubMed] [Google Scholar]
- 20.Osman KT, Maselli DB, Idilman IS Liver stiffness measured by either magnetic resonance or transient elastography is associated with liver fibrosis and is an independent predictor of outcomes among patients with primary biliary cholangitis [published online September 23, 2020]. J Clin Gastroenterol. doi:10.1097/MCG.0000000000001433. [DOI] [PMC free article] [PubMed]