factors associated with progression and outcomes of early

Clinical Gastroenterology and Hepatology 2020;18:684–692

Factors Associated With Progression and Outcomes of EarlyStage Primary Biliary Cholangitis

Nikolaos K. Gatselis,*,a Jorn C. Goet,‡,a Kalliopi Zachou,* Willem J. Lammers,‡

Harry L. A. Janssen,§ Gideon Hirschfield,k Christophe Corpechot,¶

Keith D. Lindor,#,** Pietro Invernizzi,‡‡ Marlyn J. Mayo,§§ Pier Maria Battezzati,kk

Annarosa Floreani,¶¶ Albert Pares,## Vasiliki Lygoura,*,‡‡ Frederik Nevens,***Andrew L. Mason,‡‡‡ Kris V. Kowdley,§§§ Cyriel Y. Ponsioen,kkk Tony Bruns,¶¶¶,###

Douglas Thorburn,**** Xavier Verhelst,‡‡‡‡ Maren H. Harms,‡ Henk R. van Buuren,‡

Bettina E. Hansen,‡,§,§§§§ and George N. Dalekos,* on behalf of the Global PrimaryBiliary Cholangitis Study Group

*Department of Medicine and Research Laboratory of Internal Medicine, School of Medicine, Thessaly University, Larissa,Greece; ‡Gastroenterology and Hepatology, Erasmus University Medical Center, Rotterdam, The Netherlands; §Toronto Centrefor Liver Disease, Toronto General Hospital, University of Toronto, Toronto, Canada; kNational Institute for Health Research(NIHR) Biomedical Research Unit and Centre for Liver Research, University of Birmingham, Birmingham, United Kingdom;¶Centre de Reference des Maladies Inflammatoires des Voies Biliaires, Hôpital Saint-Antoine, Paris, France; #Department ofGastroenterology and Hepatology, Mayo Clinic, Rochester, Minnesota; **Arizona State University, Phoenix, Arizona; ‡‡Divisionof Gastroenterology and Center for Autoimmune Liver Diseases, Department of Medicine and Surgery, University of MilanBicocca, Milan, Italy; §§Digestive and Liver Diseases, University of Texas Southwestern Medical Center, Dallas, Texas;kkDepartment of Health Sciences, Universita degli Studi di Milano, Milan, Italy; ¶¶Department of Surgery, Oncology andGastroenterology, University of Padua, Padua, Italy; ##Liver Unit, Hospital Clinic, Centro de Investigación Biomédica en Red deEnfermedades Hepáticas y Digestivas (CIBERehd), Institut D’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS),University of Barcelona, Barcelona, Spain; ***Department of Hepatology, University Hospitals Leuven, Katholieke Universiteit(KU) Leuven, Leuven, Belgium; ‡‡‡Division of Gastroenterology and Hepatology, University of Alberta, Edmonton, Alberta,Canada; §§§Liver Care Network, Swedish Medical Center, Seattle, Washington; kkkDepartment of Gastroenterology andHepatology, Academic Medical Center, Amsterdam, The Netherlands; ¶¶¶Department of Internal Medicine IV, Jena UniversityHospital, Friedrich Schiller University, Jena, Germany; ###Department of Medicine III, University Hospital RheinischWestfälische Technische Hochschule (RWTH) Aachen, Aachen, Germany; ****The Sheila Sherlock Liver Centre, The Royal FreeHospital, London, United Kingdom; ‡‡‡‡Department of Gastroenterology and Hepatology, Ghent University Hospital, Ghent,Belgium; §§§§Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada

BACKGROUND & AIMS:

aAuthors share co-first authors

Abbreviations used in this papetransaminase; AST, aspartateBiliary Cholangitis Study Grourange; LT, liver transplantationursodeoxycholic acid; ULN, upp

Patients usually receive a diagnosis of primary biliary cholangitis (PBC) at an early stage,based on biochemical analyses. We investigated the proportion of patients who progress tomoderate or advanced PBC and factors associated with progression and patient survival.

METHODS:
We obtained data from 1615 patients (mean age, 55.4 y) with early stage PBC (based on theirnormal levels of albumin and bilirubin), collected at the time of initial evaluation ortreatment, from the Global PBC Study Group database (comprising patients at 19 livercenters in North American and European countries). We collected data from health careevaluations on progression to moderate PBC (abnormal level of bilirubin or albumin) oradvanced-stage PBC (abnormal level of both). The median follow-up time was 7.9 years. Thecomposite end point was decompensation, hepatocellular carcinoma, liver transplantation,or death.
RESULTS:

Of the 1615 patients identified with early stage PBC, 904 developed moderate PBC and 201developed advanced disease over the study period. The proportions of patients who tran-sitioned to moderate PBC at 1, 3, and 5 years were 12.9%, 30.2%, and 45.8%. The

hip.

r: ALP, alkaline phosphatase; ALT, alaninetransaminase; GPBCSG, Global Primaryp; HR, hazard ration; IQR, interquartile; PBC, primary biliary cholangitis; UDCA,er limit of normal.

Most current article

© 2020 by the AGA Institute1542-3565/$36.00

https://doi.org/10.1016/j.cgh.2019.08.013


http://crossmark.crossref.org/dialog/?doi=10.1016/j.cgh.2019.08.013&domain=pdf

March 2020 Biochemical Transitions in PBC 685

proportions of these patients who then transitioned to advanced PBC at 1, 3, and 5 yearslater were 3.4%, 12.5%, and 16.0%, respectively. During the follow-up period, 236 patientshad a clinical event. The proportions of patients with moderate PBC and event-free survivalwere 97.9%, 95.1%, and 91.5% at 1, 3, and 5 years, respectively, and the proportions ofpatients with advanced PBC and event-free survival were 90.6%, 71.2%, and 58.3% at 1, 3,and 5 years later, respectively. Variables associated with transition from early to moderatePBC included baseline levels of bilirubin, albumin, and alkaline phosphatase; aspartate toalanine aminotransferase ratio; platelet count; and treatment with ursodeoxycholic acid.Transitions from early to moderate PBC and from moderate to advanced PBC were associ-ated with higher probabilities of a clinical event (time-dependent hazard ratios, 3.0; 95% CI,2.0–4.5; and 4.6; 95% CI, 3.5–6.2).

CONCLUSIONS:
Approximately half of patients with early stage PBC progress to a more severe stage within 5years. Progression is associated with an increased risk of a clinical event, so surveillance isimportant for patients with early stage PBC.
Keywords: Autoimmune Liver Disease; Prognostic Factor; Antimitochondrial Antibodies; UDCA.

Primary biliary cholangitis (PBC) is a slowlyprogressive cholestatic liver disease that may

lead to cirrhosis and liver failure, requiring livertransplantation (LT). Currently, ursodeoxycholic acid(UDCA) is the first-line therapy, but new therapies arebecoming available.1

The clinical course of PBC varies greatly, and itsclinical presentation has changed substantially overrecent decades. In the past, most patients were diag-nosed at symptomatic and advanced disease stages(extensive fibrosis or cirrhosis),2 but the majority ofPBC patients seen in recent clinical practice are diag-nosed at asymptomatic and earlier stages.3–5 This shiftin clinical presentation may be owing to higher diseaseawareness,1 improved diagnostic assays for anti-mitochondrial antibody detection,6 and/or moreroutine testing of liver function tests. Regardless, giventhe shift, relying solely on hard clinical end points,such as death or LT, may not be feasible in clinicalstudies of patients with PBC.7

The long-term survival of PBC patients with anearly biochemical stage (defined as normal albuminand normal bilirubin levels based on the Rotterdamcriteria) generally is comparable with the survival rateof the general population.8 These patients thereforeare perceived as having a low risk of developing pro-gressive disease. However, patients identified as lowrisk at the beginning of the disease still may progressto a moderate (abnormal albumin or bilirubin levels)or advanced stage (both abnormal albumin and bili-rubin levels) during follow-up evaluation, which maybe associated with worse long-term survival. Knowl-edge regarding predictive factors for biochemicaltransitions, as well as their impact on prognosis, mayaid in identifying patients who are likely to progressover time. Therefore, in a cohort of PBC patients withearly biochemical stage, we examined the proportionwho progressed to moderate or advanced PBC andfactors associated with progression and patientsurvival.

Methods

Patients

This study was a subgroup analysis of patientsincluded in the Global PBC Study Group (GPBCSG)database, a multicenter collaboration between 19 livercenters from 12 North American and European coun-tries. All patients had an established diagnosis of PBC.1

Follow-up data were collected prospectively.For the current study, only those patients with bio-

chemically early disease at baseline according to theRotterdam criteria9 were included and evaluated fortransition to biochemically moderate and/or advanceddisease during follow-up evaluation. Both UDCA-treatedand untreated patients were included. Exclusioncriteria and collected data are listed in theSupplementary Methods section.

This study was conducted in accordance with theprotocol and principles of the 1975 Declaration of Hel-sinki. The protocol was approved by the InstitutionalResearch Board of the corresponding center, and at eachparticipating center, in accordance with local regulations.

End Points

The end points included progression to moderate PBC(abnormal level of bilirubin or albumin) or advanced-stage PBC (abnormal level of both bilirubin and albu-min) and progression to the composite clinical end point,defined by either LT, ascites, variceal bleeding, hepaticencephalopathy, development of hepatocellular carci-noma, or death, whichever occurred first.

Statistical Analysis

The baseline visit was considered as the start date ofUDCA or the date of initial evaluation in case of untreatedpatients. A competing-risk approach was used: patients

What You Need to Know

BackgroundPatients usually receive a diagnosis of primarybiliary cholangitis (PBC) at an early stage. We stud-ied the proportion of patients who progress tomoderate or advanced PBC, factors associated withprogression, and patient survival.

FindingsApproximately half of patients with early stage PBCprogress to a more severe stage within 5 years.However, most patients with moderate or advancedPBC survive until this time point. We identifiedbiochemical and treatment factors associated withthe progression of PBC.

Implications for patient careThe progression of PBC is associated with anincreased risk of a clinical event, so surveillance isimportant for patients with early stage PBC.

686 Gatselis et al Clinical Gastroenterology and Hepatology Vol. 18, No. 3

who remained stable at biochemical early disease duringfollow-up evaluation or patients who did not reach anyof the components of the composite clinical event werecensored at their last follow-up visit. To determine theoverall impact of transitions between biochemical stageson event-free survival, the time until patients transi-tioned from early to moderate stage, and early toadvanced biochemical stage, was modeled as a time-dependent covariate.

Biochemical data were not available for all patients.Missing data varied between 8% for alkaline phosphatase(ALP) and 26.6% for aspartate aminotransferase (AST)values; it is probable that stable patients had fewer lab-oratory measurements, contrary to the closer follow-upevaluation of patients with an accelerated progression,thereby creating a bias in the availability of data. Becauseof the extensive data collection effort for the GPBCSG,laboratory data were available for a total of 75,000 visitsbetween UDCA initiation/date of initial evaluation untilthe end of follow-up evaluation or the occurrence of aclinical event. As such, the trajectory of the laboratoryvalues over time was used as prior information for theimputation of missing values as well as the strong cor-relation between multiple different laboratory values andthe end point. Therefore, we used an imputed data set forour primary analysis to ensure full detection ofbiochemical transitions. In detail, SAS (SAS multipleimputation procedure, Markov Chain Monte Carlo(MCMC) method; SAS software, version 9.3, SAS Institute,Cary, NC) was used to generate 10 imputed data sets oflaboratory results at yearly time points between the startof UDCA treatment or the initial evaluation and up to 15years of follow-up evaluation. Missing data were consid-ered as missed at random. Rubin’s rule10,11 rules wereused for estimation of the parameters and the standarderror. The imputation model included baseline variablesthat were potentially predictive of PBC outcomes (eg, yearof diagnosis, age) as well as the outcomes themselves.Only continuous biochemical variables were imputed.

Univariate and multivariate Cox regression analyseswere performed to assess the impact of various factorson the rate of biochemical transition from biochemicallyearly to moderate disease, and from moderate toadvanced disease. In the analyses of factors associatedwith transition from moderate to advanced disease,laboratory parameters correspond to the time point ofmoderate disease development. In multivariable ana-lyses, the model with the lowest Akaike InformationCriteria was chosen. The effect of albumin and bilirubinwithin the normal range are presented on a continuousscale and by a binary split at the median for reasons ofclinical interpretation and the aim of comparing groupsof equal sample size. In addition, transition rates wereassessed and compared between patients with baselineGLOBE scores above the age-specific GLOBE scorethreshold, and those remaining below this thresholdusing Kaplan–Meier estimates. The calculation and use ofGLOBE score thresholds have been described previously,

with patients above the threshold presenting a signifi-cantly worse survival compared with matched in-dividuals from the general population.12

Normally distributed data are presented as means �SD, and skewed distributed data are presented as medianand interquartile range (IQR). Where indicated, contin-uous variables underwent natural logarithmic trans-formation to correct for nonlinearity. All analyses were2-sided. A P value less than .05 was considered statisti-cally significant. Statistics analyses were performed withIBM SPSS Statistics 22.0 (SPSS, Inc, Chicago, IL).

Results

Study Population Characteristics

The patient population consisted of 1615 patients withbiochemically early PBC (Supplementary Figure 1). Themean age at study entry was 55.4 years (SD, 11.9 y). Themedian total follow-upperiodwas7.9 years (IQR, 4.3–12.5y; range, 0.5–25 y). Baseline patient characteristics areshown in Table 1. The majority of patients (1415 of 1615;87.6%) were treated with UDCA. Histologic disease stage(according to the Ludwig et al13 and Scheuer14 classifica-tion) was available for 798 (49%) patients, of whommosthad early disease (stages I or II, 623 of 798; 78%).

Of the 1415 UDCA-treated patients, a total of 1383 hada follow-up period that was longer than 1 year, withbiochemical nonresponse rates, according to variouscriteria, ranging from 14.0% with the Paris I criteria to59.9%with theBarcelona criteria. According to theGLOBEscore, only 11.0% of patients presenting with early stagePBC had an estimated survival rate that was worse thanthat of the general population at 1 year of UDCA therapy.

Table 1. Baseline Study Population Characteristics

Total cohort (N ¼ 1615)

Age at diagnosis, y 53.4 (12.0)Age at study entry, y 55.4 (11.9)Year of diagnosis, median (IQR) 1997 (1990–2004)Year of diagnosis, range 1961–2014Female, n (%) 1480 (91.6)AMAþ, n (%) 1459 (90.3)UDCA treated, n (%) 1415 (87.6)Biopsy stage,a n (%)

I 381 (47.7)II 242 (30.3)III 119 (14.9)IV 56 (7.0)

Serum total bilirubin level, �ULN 0.52 (0.40–0.70)Serum albumin level, �LLN 1.19 (1.11–1.26)Serum ALP level, �ULN 1.93 (1.25–3.11)Serum AST level, �ULN 1.28 (0.90–1.87)Serum ALT level, �ULN 1.50 (0.95–2.33)Serum platelets, �103/mm3 255 (203–366)

NOTE. Data are represented as means (SD) and medians (IQR).ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA, anti-mitochondrial antibodies; AST, aspartate aminotransferase; IQR, interquartilerange; LLN, lower limit of normal; UDCA, ursodeoxycholic acid; ULN, upperlimit of normal.aBaseline biopsy specimens (obtained within 1 year of start of UDCA) wereavailable in 798 of 1615 (49%) patients. Histologic stage is determinedaccording to the Ludwig et al13 and Scheuer14 classification.


Transitions in Biochemical Stage

During the median of 7.9 years (IQR, 4.3–12.5 y) offollow-up evaluation, 904 patients transitioned frombiochemically early to moderately advanced stage.

Figure 1. Transitions of patients according to their biochemicalfour patients transitioned directly to advanced stage. For thesmoderate to advanced stages. bEight cirrhotic decompensatiocirrhotic decompensations, 11 HCC, 7 LTx, and 47 deaths (7 live34 deaths (9 liver-related). HCC, hepatocellular carcinoma, LTx

Sequentially, 201 of 904 patients transitioned frommoderate to advanced disease (Figure 1). Thirty-fourpatients transitioned directly from early to advancedstage. For these patients, we assumed that the transitiontook place via moderate stage. Overall, the transitionrates from early to moderate stage were 12.9% at 1 year,30.2% at 3 years, and 45.8% at 5 years of follow-upevaluation (Figure 2A). The cumulative transition ratesfrom moderate to advanced (201 of 904) stage were3.4% at 1 year, 12.5% at 3 years, and 16.0% at 5 years offollow-up evaluation (Figure 2B). The median time totransition to moderate stage was 2.5 years (IQR, 1–5 y;range, 0.5–15 y). The median time from moderate toadvanced stage was 1.5 years (IQR, 0.5–3.5 y; range,0.5–11.5 y).

Factors Associated With Transition FromBiochemical Early to Moderately AdvancedDisease Stage

In univariate analyses, baseline factors associatedwith progression from biochemically early to moderatedisease were male sex, older age at study entry, bilirubinlevel, albumin level, ALP level, and AST/alanine amino-transferase (ALT) ratio. Patients with a more recentdiagnosis and higher platelet levels at baseline and wereless likely to transition to moderate stage (Table 2). Bi-opsy (available for 798 patients) stages III and IV wereassociated with a higher probability of disease progres-sion compared with stage I (hazard ratio [HR], 1.7; 95%CI, 1.3–2.2; P < .001; and HR, 1.8; 95% CI, 1.3–2.5; P <.001, respectively), while stage II was not associated with

stage during follow-up evaluation and clinical events. aThirty-e patients, we assumed they transitioned gradually throughns, no HCC, no LTx, and 31 deaths (5 liver-related). cFortyr-related). dThirty-nine decompensations, 8 HCC, 11 LTx, and, liver transplantation.

Figure 2. Cumulative incidence of biochemical transition andevents. Kaplan–Meier estimates of (A) transition from mild toa moderate stage or event, and (B) transition from a moderateto advanced disease stage or an event.


a higher probability of disease progression (P ¼ .14). Forpatients with histologic stages I or II, the transition ratesfrom early to moderate stage at 1, 3, and 5 year(s) were10.9%, 31.5%, and 43.6%, respectively. The transitionrates were higher for those with stage III or IV: 18.4%,41.5%, and 60.2% at 1, 3, and 5 year(s), respectively.

Table 2. Baseline Factors Associated With the Transition From

Univariate

HR 95%

Male sex 1.27 1.01–Age at entry, per 10 years 1.06 1.00–Year of diagnosis, per decade 0.86 0.78–UDCA use 0.90 0.73–Bilirubin, �ULNa 2.14 1.78–Albumin, �LLN 0.07 0.04–ALP, �ULNa 1.40 1.28–AST/ALT ratioa 1.33 1.11–Platelets, per 10-unit increase, �103/mm3 0.98 0.97–

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotracid; ULN, upper limit of normal.aThese biochemical variables were transformed with a natural logarithm.

In multivariable analyses, all variables except age atstudy entry and year of diagnosis remained significantlyassociated with transition to moderate stage (Table 2).Bilirubin and albumin at a binary split of >0.5 times theupper limit of normal (ULN) and �1.2 times the ULN,respectively, were associated significantly with biochem-ical transition in multivariable analyses (HR, 1.5; 95% CI,1.3–1.8; P < .001; and HR, 1.6; 95% CI, 1.4–1.9; P < .001,respectively). The baseline AST/ALT ratio (HR, 1.3; 95%CI, 1.1–1.5) and ALP levels (HR, 1.3; 95% CI, 1.2–1.5) werepositive predictors of progression, although patientstreated with UDCA had a lower transition rate (HR, 0.70;95% CI, 0.57–0.86). In addition, ALP levels greater than1.67 times the ULN was associated with transition tomoderate stage (HR, 1.4; 95%CI, 1.2–1.6) (SupplementaryFigure 2). Bilirubin level, albumin level, ALP level, AST/ALT ratio, and platelets retained their prognostic value forbiochemical transition in the subgroup of UDCA-treatedpatients (Supplementary Table 1).

A total of 242 of 1615 patients with normal albuminand bilirubin levels at baseline had GLOBE score valuesbeyond the age-specific GLOBE score threshold. Thesepatients were more likely to progress to moderate stage(HR, 1.8; 95% CI, 1.6–2.1) (Supplementary Figure 3).

In subgroup analyses including only histologicallyproven early stage patients (stages, I–II), all factorsexcept platelets and AST/ALT ratio remained as inde-pendent predictors of transition from early to moderatebiochemical stage (Supplementary Table 2).

Factors Associated With Transition FromBiochemical Moderately Advanced toAdvanced Disease Stage

In 904 patients who transitioned to the moderatestage, univariate analyses showed older age, total bili-rubin level, albumin level, ALP level, and aminotrans-ferase levels at the time point of transition to moderatestage as predictive factors for subsequent transition to

Biochemically Early to Moderately Advanced Disease

analyses Multivariate analysesa

CI P HR 95% CI P

1.60 .043 1.27 1.01–1.60 .0451.12 .044 – – –

0.94 <.001 – – –

1.09 .279 0.70 0.57–0.86 .0012.58 <.001 1.86 1.53–2.26 <.0010.12 <.001 0.08 0.04–0.16 <.0011.55 <.001 1.33 1.20–1.48 <.0011.58 .002 1.27 1.06–1.53 <.0010.99 <.001 0.99 0.98–0.99 .007

ansferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholic


advanced disease (Table 3). In contrast, UDCA-treatedpatients and patients with higher platelet levels wereless likely to transition to advanced disease. Althoughmultivariate analyses rendered age and plateletsnonsignificant, all other variables remained associatedwith biochemical transition to advanced stage, withUDCA associated with a lower probability of transition(HR, 0.57; 95% CI, 0.40–0.82). ALP levels greater than1.67 times the ULN were associated significantly withtransition to the advanced stage (HR, 2.5; 95% CI,1.9–3.3) (Supplementary Figure 2). Albumin levels (bi-nary split of <1 times the lower limit of normal) werenot associated with biochemical transition whereas bili-rubin levels (at a binary split of >1 times the ULN; HR,2.0; 95% CI, 1.4–1.7) were associated with transition tothe advanced stage. Patients with GLOBE score valueshigher than the threshold (568 of 904) had higher ratesof progression to advanced disease than those withGLOBE scores lower than the threshold (HR, 3.0; 95% CI,2.0–4.3) (Supplementary Figure 3).

Bilirubin level, albumin level, ALP level, and AST/ALTratio also were associated with biochemical transition tothe advanced stage in the subgroup of UDCA-treatedpatients (Supplementary Table 3).

The Impact of Biochemical Transition on Events

During follow-up evaluation, 236 patients developedat least 1 clinical event: 87 decompensations, 19hepatocellular carcinomas, 18 LTs, and 112 deaths (21liver-related deaths) (Figure 1). Overall, the event-freesurvival for patients remaining in early biochemicaldisease stage at 1, 3, 5, and 10 years was 99.8%, 98.9%,98.3%, and 93.2%, respectively (Figure 3). Once patientsreached a moderate stage (n ¼ 904), the event-freesurvival rates at 1, 3, and 5 year(s) of follow-up evalu-ation were 97.9%, 95.1%, and 91.5%, respectively. Forpatients who consequently transitioned to an advanced

Table 3. Factors Associated With the Transition From Biochem

Univariate

HR 95%

Male sex 1.22 0.78–1Age, per 10 yearsa 1.23 1.11–1Year of diagnosis, per decadeb 0.86 0.69–1UDCA use 0.45 0.31–0Bilirubin, �ULNb,c 2.69 2.09–3Albumin, �LLN 0.36 0.15–0ALP, �ULNb,c 2.21 1.80–2AST/ALT ratiob,c 1.51 1.08–2Platelets, per 10-unit increase, �103/mm3 0.98 0.96–0

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotracid; ULN, upper limit of normal.aAge at which biochemical moderate stage was reached.bThese biochemical variables were transformed with a natural logarithm.cLaboratory parameters correspond to the time point of moderate disease develo

biochemical stage (n ¼ 201), these rates were 90.6%,71.2%, and 58.3%, respectively (Figure 3).

In univariate and multivariate Cox regression, inwhich biochemical transition was modeled as a time-dependent covariate, a higher probability of a clinicalevent during follow-up evaluation was found for patientswho transitioned from a biochemical early to moderatestage (HR, 3.0; 95% CI, 2.0–4.5), from a biochemicalmoderate to advanced stage (HR, 4.6; 95% CI, 3.5–6.2),and overall from an early to advanced disease stage (HR,14.1; 95% CI, 9.3–21.4; P < .001) (SupplementaryTable 4).

Discussion

In this study of a large cohort of more than 1500 PBCpatients from European and North American centers, wefound that approximately 1 of 2 patients presenting withan early biochemical disease stage transitioned toward amoderate biochemical stage within 5 years of follow-upevaluation. Almost 1 in 6 of these patients who reacheda moderate stage eventually transitioned to an advancedstage within the next 5 years. These transitions wereassociated with an increased risk of clinical events,underlining the importance of clinical surveillance evenin early stage PBC patients.

We also were able to identify the patients who aremore likely to transition into moderately advanced andadvanced stages. UDCA treatment was associated withlower rates of biochemical transition (as well as thecomposite end point) during follow-up evaluation.Extended data from several studies have indicated thatUDCA improves the natural history of PBC, even whenadministered in the early stages.3,4,8 The GPBCSGrecently showed that UDCA-treated patients have alower risk of LT or death than untreated patients, and,more importantly, the benefit was observed in both re-sponders and nonresponders.15 Our data point in the

ically Moderately Advanced to Advanced Disease (n ¼ 904)

analyses Multivariable analyses

CI P HR 95% CI P

.92 .390 – – –

.35 .001 – – –

.05 .119 – – –

.63 <.001 0.57 0.40–0.82 .002

.47 <.001 3.87 3.04–4.94 <.001

.94 .036 0.02 0.01–0.05 < .001

.71 <.001 2.05 1.65–2.54 < .001

.12 .016 1.66 1.17–2.35 .004

.99 <.001 – – –

ansferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholic

pment.

Figure 3. Event-free sur-vival in a clock-resetapproach in patients whotransit from early, to mod-erate, to advancedbiochemical diseasestage. Patients who didnot transit in biochemicalstage remain in line A.Patients who transitionedfrom early to moderatestage were switched to anew survival curve in lineB, which then was reset astime 0 for their furtherfollow-up evaluation. Pa-tients who then transi-tioned from moderate toadvanced stages wereswitched to survival curvein line C. All hazard ratios(HR) were obtained byconsidering biochemicaltransition as a time-dependent covariate inCox regression analyses.


same direction, by showing UDCA treatment in patientsat an early biochemical stage may reduce progression tomore advanced stages. Taking into account that UDCA isrecommended for all PBC patients including those atearly stages,1 we conducted a subanalysis including onlyUDCA-treated patients with similar results in regard tokey risk factors (Supplementary Tables 1 and 3).

Similar to findings in previous studies, ALP levels wereindicative of disease progression. In PBC, ALP is consid-ered one of themost robustly validatedmarkers of diseaseactivity. In meta-analyses of 4845 patients, we previouslyshowed that ALP (>2.0 times the ULN) was associatedstrongly with LT and death across various subgroups.16

Moreover, ALP is an important component of mostbiochemical criteria that assess treatment response toUDCA after 1 year of therapy. Our results indicate thathigher ALP levels are associated not onlywith hard clinicalend points, but also with biochemical transition.

In accordance with previous studies, we foundplatelet count and the AST/ALT ratio to be associatedwith biochemical disease progression, particularly inpatients with advanced histologic stages (III–IV).12,17–22

Platelet count generally is considered a marker of por-tal hypertension and is of particular importance indiscriminating noncirrhotic from cirrhotic patients withnormal bilirubin and albumin levels. Our resultsemphasize the importance of platelet count and the AST/ALT ratio in identifying PBC patients who are likely toprogress.

In contrast to previous studies, male sex was notassociated consistently with disease progression.23–25

One possible explanation for this is that only patientswith mild PBC were included in our study, whereas in aprevious study, the negative impact of male sex appearedto be limited only to patients with advanced disease.23

Alternatively, the use of the AST/ALT ratio in ourmultivariable analyses could have rendered male sex anonsignificant factor. Prior studies have documentedrelatively higher rates of alcohol consumption in men,which is characterized by an increased AST/ALT ra-tio.4,26 This finding coincides with a recent Greek clinicalstudy, in which, after adjusting for other confoundingfactors including alcohol consumption, male sex did notindependently pose a greater risk for disease progres-sion during follow-up evaluation.27 Of note, in accor-dance with our findings, a recent study from GPBCSG didnot associate sex with response and transplant-freesurvival.28

Strengths of our study were the inclusion of a largecohort of PBC patients from different geographic areas aswell as long-term follow-up evaluation with many clin-ical events. This increases the reliability and generaliz-ability of our results. The GPBCSG database captured allpatient visits and represents the current clinical prac-tices in the participating centers and the results there-fore are highly relevant to clinicians working with PBCpatients. Although few laboratory data were missing, theresults presented are based on the imputed databases.


To support our findings, a sensitivity analysis thatexcluded patients with missing data yielded similar re-sults (Supplementary Tables 5 and 6). No data on co-factors (eg, alcohol consumption or obesity) wereavailable in the GPBCSG database.8,26 However, patientswith alcoholic liver disease were excluded. Nevertheless,future studies taking into account the aforementionedcofactors are warranted to identify patient groups with ahigh risk of disease progression.

A question that remains is how often we shouldperform laboratory examinations in early PBC patients.Personalized risk stratification, using biochemicalresponse markers or prognostic models after 1 year ofUDCA therapy, can identify patients at risk of progressivedisease.1 European Association for the Study of the Liverclinical practice guidelines for PBC recommend that allpatients should have life-long follow-up evaluation,recognizing that patients have different disease coursesand may require varied levels of attention.1 Patients witha higher baseline GLOBE score had higher biochemicaltransition rates. A baseline GLOBE score calculationtherefore can aid in the identification of patients whoneed closer surveillance, even those with earlybiochemical stage. Our results also indicate that transi-tions in biochemical stage are important markers ofdisease progression that could be incorporated intodisease staging. Transitions to moderately advanced andadvanced disease stage may justify a change in follow-upregimen with closer surveillance.

In conclusion, our internationally representativestudy provides a comprehensive overview of the naturalhistory of PBC patients with early disease stage, showingthat almost 1 of 2 patients with early biochemical diseasewill transition to moderately advanced disease, andapproximately one sixth of them will progress to theadvanced stage. These transitions are associated with anincreased probability of clinical events. The findingsunderline the importance of clinical surveillance in PBCpatients with early biochemical disease stage.

Supplementary Material

Note: To access the supplementary material accom-panying this article, visit the online version of ClinicalGastroenterology and Hepatology at www.cghjournal.org,and at https://doi.org/10.1016/j.cgh.2019.08.013.

References
1. EASL Clinical Practice Guidelines: The diagnosis and manage-
ment of patients with primary biliary cholangitis. J Hepatol 2017;67:145–172.

2. Prince M, Chetwynd A, Newman W, et al. Survival and symptomprogression in a geographically based cohort of patients withprimary biliary cirrhosis: follow-up for up to 28 years. Gastro-enterology 2002;123:1044–1051.

3. Floreani A, Caroli D, Variola A, et al. A 35-year follow-up of alarge cohort of patients with primary biliary cirrhosis seen at asingle centre. Liver Int 2011;31:361–368.

4. Gatselis NK, Zachou K, Lygoura V, et al. Geoepidemiology,clinical manifestations and outcome of primary biliary chol-angitis in Greece. Eur J Intern Med 2017;42:81–88.

5. Murillo Perez CF, Goet JC, Lammers WJ, et al. Milder diseasestage in patients with primary biliary cholangitis over a 44-yearperiod: a changing natural history. Hepatology 2018;67:1920–1930.

6. Gatselis NK, Dalekos GN. Molecular diagnostic testing for pri-mary biliary cholangitis. Exp Rev Mol Diagn 2016;16:1001–1010.

7. Prince MI, James OF. The epidemiology of primary biliarycirrhosis. Clin Liver Dis 2003;7:795–819.

8. ter Borg PC, Schalm SW, Hansen BE, et al. Prognosis ofursodeoxycholic acid-treated patients with primary biliarycirrhosis. Results of a 10-yr cohort study involving 297 patients.Am J Gastroenterol 2006;101:2044–2050.

9. Kuiper EM, Hansen BE, de Vries RA, et al. Improved prognosisof patients with primary biliary cirrhosis that have a biochemicalresponse to ursodeoxycholic acid. Gastroenterology 2009;136:1281–1287.

10. Little RJA, Rubin DB. Statistical analysis with missing data. NewYork: John Wiley & Sons, 1987.

11. Rubin D. Multiple imputation after 18þ years. J Am Stat Assoc1996;91:473–489.

12. Lammers WJ, Hirschfield GM, Corpechot C, et al. Developmentand validation of a scoring system to predict outcomes of pa-tients with primary biliary cirrhosis receiving ursodeoxycholicacid therapy. Gastroenterology 2015;149:1804–1812.

13. Ludwig J, Dickson ER, McDonald GS. Staging of chronic non-suppurative destructive cholangitis (syndrome of primary biliarycirrhosis). Virchows Arch A Pathol Anat Histol 1978;379:103–112.

14. Scheuer P. Primary biliary cirrhosis. Proc R Soc Med 1967;60:1257–1260.

15. van der Meer AJ, Harms MH, Corpechot C, et al. Ursodeox-ycholic acid is associated with a prolonged transplant-freesurvival in all patients with primary biliary cholangitis - there isno such thing as non-response. Hepatology 2017;66:155A.

16. Lammers WJ, van Buuren HR, Hirschfield GM, et al. Levels ofalkaline phosphatase and bilirubin are surrogate end points ofoutcomes of patients with primary biliary cirrhosis: an interna-tional follow-up study. Gastroenterology 2014;147:1338–1349.

17. Trivedi PJ, Bruns T, Cheung A, et al. Optimising risk stratificationin primary biliary cirrhosis: AST/platelet ratio index predictsoutcome independent of ursodeoxycholic acid response.J Hepatol 2014;60:1249–1258.

18. Joshita S, Umemura T, Ota M, et al. AST/platelet ratio indexassociates with progression to hepatic failure and correlateswith histological fibrosis stage in Japanese patients with primarybiliary cirrhosis. J Hepatol 2014;61:1443–1445.

19. Patanwala I, McMeekin P, Walters R, et al. A validated clinicaltool for the prediction of varices in PBC: the Newcastle Varicesin PBC Score. J Hepatol 2013;59:327–335.

20. Trivedi PJ, Lammers WJ, van Buuren HR, et al. Stratification ofhepatocellular carcinoma risk in primary biliary cirrhosis: amulticentre international study. Gut 2016;65:321–329.

21. Nyblom H, Björnsson E, Simrén M, et al. The AST/ALT ratio asan indicator of cirrhosis in patients with PBC. Liver Int 2006;26:840–845.

22. Reig A, Pares A. Lowplatelet count influences the performance ofthe different biochemical criteria of ursodeoxycholic acid therapyresponse in primary biliary cholangitis. J Hepatol 2017;66:S355.

http://www.cghjournal.org


http://refhub.elsevier.com/S1542-3565(19)30888-2/sref1




























































































23. Trivedi PJ, Corpechot C, Pares A, et al. Risk stratification inautoimmune cholestatic liver diseases: opportunities for clini-cians and trialists. Hepatology 2016;63:644–659.

24. Carbone M, Mells GF, Pells G, et al. Sex and age are de-terminants of the clinical phenotype of primary biliary cirrhosisand response to ursodeoxycholic acid. Gastroenterology 2013;144:560–569.

25. Cheung A, Lammers WJ, Hirschfield GM, et al. Age, bilirubin andalbumin, regardless of sex, are the strongest independent pre-dictors of biochemical response and transplantation-free sur-vival in patients with primary biliary cirrhosis. J Hepatol 2015;62:S798–S799.

26. Harada K, Hirohara J, Ueno Y, et al. Incidence of and risk factorsfor hepatocellular carcinoma in primary biliary cirrhosis: nationaldata from Japan. Hepatology 2013;57:1942–1949.

27. Azariadis K, Lygoura V, Zachou K, et al. Characteristics of Greekmale patients with primary biliary cholangitis. J Hepatol 2017;66:S552.

28. Cheung AC, Lammers WJ, Murillo Perez CF, et al. Effects of ageand sex of response to ursodeoxycholic acid and transplant-freesurvival in patients with primary biliary cholangitis. Clin Gastro-enterol Hepatol 2019;17:2076–2084.e2.

Reprint requestsAddress requests for reprints to: Nikolaos K. Gatselis, MD, PhD, Department ofMedicine and Research Laboratory of Internal Medicine, Medical School,University of Thessaly, Panepistimiou 3, Biopolis, Larissa 41500, Greece.e-mail: [email protected]; fax: (30) 2413501557.

Conflicts of interestThese authors disclose the following: Tony Bruns has received honoraria fromIntercept Pharmaceuticals, Falk, and Norgine, and travel expenses from Gileadand Intercept Pharmaceuticals; Henk R. van Buuren received unrestrictedgrants from Intercept Pharmaceuticals; Christophe Corpechot is a consultantfor Intercept Pharmaceuticals and Inventiva Pharma, receives grants fromIntercept Pharmaceuticals and Arrow Generiques, and has lectured for Glax-oSmithKline; George N. Dalekos has served as an advisor/lecturer for AbbVie,Bristol-Myers Squibb, Gilead, Novartis, Bayer, and Roche, and has receivedgrant support from Bristol-Myers Squibb, Gilead, and Roche; Annarosa Flor-eani has acted as advisor to the Primary Biliary Cholangitis Committee,sponsored by Intercept; Nikolaos K. Gatselis has served as an advisor/lecturerfor AbbVie, Bristol-Myers Squibb, and Gilead; Bettina E. Hansen has receivedunrestricted grants from Intercept Pharmaceuticals, and is a consultant forIntercept Pharmaceuticals, CymaBay, and Janssen Pharma; Maren H. Harmshas received a speaker’s fee from Zambon; Gideon Hirschfield has acted as anadvisor/consultant for GlaxoSmithKline, Novartis, CymaBay Therapeutics,Intercept Pharmaceuticals, and Falk; Pietro Invernizzi has received a speaker’sfee from Zambon; Harry L. A. Janssen has received grants and acted as aconsultant for AbbVie, Bristol Myers Squibb, Gilead Sciences, Janssen, Inc,Medimmune, Merck, Roche, Arbutus, and Vir Biotechnology; Andrew L. Masonis a consultant for Intercept Pharmaceuticals and has received research sup-port from Intercept Pharmaceuticals, Merck Canada, AbbVie, and Gilead Sci-ences; Marlyn J. Mayo has received research grants from InterceptPharmaceuticals; Albert Pares has acted as an advisor for Intercept Pharma-ceuticals and Novartis; Cyriel Y. Ponsioen has received grant support fromTakeda, speaker’s fees from AbbVie, Takeda, and Dr Falk Pharma, and servedas a consultant for Takeda; and Xavier Verhelst has served as a consultant forIntercept Pharmaceuticals, Gilead, MSD, and AbbVie, and received speaker’sfees from Gilead, Bayer, MSD, and AbbVie. The remaining authors disclose noconflicts.

FundingThis investigator-initiated study was supported by unrestricted grants fromIntercept Pharmaceuticals and was funded by the Foundation for Liver andGastrointestinal Research (a not-for-profit foundation) in Rotterdam, TheNetherlands. The supporting parties had no influence on the study design, datacollection and analyses, writing of the manuscript, or the decision to submit themanuscript for publication.




























mailto:[email protected]

March 2020 Biochemical Transitions in PBC 692.e1

Supplementary Methods

Exclusion Criteria, and Collected Clinical,Laboratory, and Histologic Data

Early stage was defined by normal bilirubin andalbumin levels, moderately advanced disease wasdefined by an abnormal bilirubin or albumin level, andadvanced disease was defined by abnormal bilirubinand albumin levels. Exclusion criteria were missinglaboratory data at baseline, unknown start date oftreatment with UDCA and/or last follow-up date, shortfollow-up period (<6 months or only baseline visitavailable), or a concomitant liver disease includingautoimmune hepatitis/PBC variant and alcoholic liverdisease.

Collected clinical and laboratory data included sex,age, diagnosis of PBC, liver histology, treatment (type ofmedication, dosage, and duration), duration and last date

of follow-up evaluation, baseline antimitochondrialantibody status, laboratory values (serum ALP, totalbilirubin, albumin, AST, ALT, and platelets), and out-comes (death and cause of death, liver transplantation,hepatocellular carcinoma, ascites, variceal bleeding, andhepatic encephalopathy).

Liver histology performed within 1 year of studyentry or documented cirrhosis before study entry wasclassified as a baseline biopsy. Histologic data wereassessed for severity according to the Ludwig et al1 andScheuer2 classification.

Supplementary References
1. Ludwig J, Dickson ER, McDonald GS. Staging of chronic non-
suppurative destructive cholangitis (syndrome of primary biliarycirrhosis). Virchows Arch A Pathol Anat Histol 1978;379:103–112.

2. Scheuer P. Primary biliary cirrhosis. Proc R Soc Med 1967;60:1257–1260.









Supplementary Figure 2. Cumulative incidence of biochemical transition and events stratified by ALP level. Kaplan–Meierestimates of (A and B) transition from an early to a moderately advanced disease stage or event, and (C and D) transitionfrom a moderately advanced to advanced disease stage or an event, stratified according to ALP level. ALP, alkaline phos-phatase; ULN, upper limit of normal.

SupplementaryFigure 1. Flow chart ofpatients included in thestudy. UDCA, ursodeox-ycholic acid;

692.e2 Gatselis et al Clinical Gastroenterology and Hepatology Vol. 18, No. 3

Supplementary Figure 3. Cumulative incidence of biochemical transition and events stratified by GLOBE score. Kaplan–Meierestimates of (A and B) transition from an early to a moderately advanced disease stage or event, and (C and D) transition from amoderately advanced to advanced disease stage or an event, stratified according to GLOBE score.


Supplementary Table 1. Baseline Factors Associated With the Transition From Biochemically Early to Moderately AdvancedDisease in UDCA-Treated Patients Only (n ¼ 1415)

Univariate analyses Multivariable analysesa

HR 95% CI P HR 95% CI P

Male sex 1.33 1.05–1.70 .02 1.28 0.99–1.64 .057Age at entry, per 10 years 1.04 0.98–1.11 .176 – – –

Year of diagnosis, per decade 0.88 0.80–0.97 <.01 – – –

Bilirubin level, �ULNa 2.05 1.69–2.50 <.001 1.82 1.48–2.25 <.001Albumin level, �LLN 0.07 0.04–0.14 <.001 0.10 0.05–0.20 <.001ALP level, �ULNa 1.33 1.20–1.48 <.001 1.30 1.16–1.46 <.001AST/ALT ratioa 1.33 1.11–1.61 <.01 1.31 1.08–1.60 .006Platelets, per 10-unit increase, �103/mm3 0.99 0.99–0.99 <.01 0.99 0.98–0.99 .031

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; ULN, upper normal limitof normal.aThese biochemical variables were transformed with a natural logarithm.

Supplementary Table 2. Baseline Factors Associated With the Transition From Biochemically Early to Moderately AdvancedDisease in Patients With Biopsy Stages I or II (n ¼ 623)



Male sex 1.30 0.91–1.85 .153 – – –

Age at entry, per 10 years 1.06 0.96–1.16 .232 – – –

Year of diagnosis, per decade 0.81 0.68–0.94 .002 0.89 0.75–1.03 .098UDCA usea 0.93 0.59–1.48 .772 – – –

Bilirubin level, �ULNb 1.94 1.43–2.62 <.001 1.75 1.28–1.38 <.001Albumin level, �LLN 0.12 0.04–0.38 <.001 0.14 0.05–0.41 <.001ALP level, �ULNb 1.35 1.15–1.58 <.001 1.23 1.05–1.45 .012AST/ALT ratiob 1.21 0.90–1.61 .208 – – –

Platelets, per 10-unit increase, �103/mm3 0.99 0.98–1.01 .366 – – –

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholicacid; ULN, upper normal limit of normal.aUDCA was not entered into the multivariable analysis because only 34 patients were not treated with UDCA.bThese biochemical variables were transformed with a natural logarithm.


Supplementary Table 3. Factors Associated With the Transition From Biochemically Moderately Advanced to AdvancedDisease in UDCA-Treated Patients Only (n ¼ 794)



Male sex 1.38 0.85–2.22 .192 – – –

Age at entry, per 10 yearsa 1.19 1.06–1.33 .011 – – –

Year of diagnosis, per decade 0.89 0.68–1.10 .276 – – –

Bilirubin level, �ULNb,c 2.57 1.95–3.40 <.001 3.74 2.88–4.85 <.001Albumin level, �LLNc 0.39 0.14–1.10 .076 0.02 0.01–0.06 <.001ALP level, �ULNb,c 1.94 1.54–2.44 <.001 1.98 1.54–2.53 <.001AST/ALT ratiob,c 1.43 0.98–2.09 .067 1.72 1.14–2.59 .004Platelets, per 10-unit increase, �103/mm3 0.99 0.99–1.00 .239 – – –

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholicacid; ULN, upper normal limit of normal.aAge at which biochemical moderate stage was reached.bThese biochemical variables were transformed with a natural logarithm.cLaboratory parameters correspond to the time point of moderate disease development.

Supplementary Table 4. The Association of Time-Dependent Transition With Clinical Events

Univariate analyses Multivariable analyses


Moderate vs early biochemical stagea 3.37 2.28–4.97 <.001 3.05 2.06–4.51 <.001Advanced vs moderate biochemical stagea 5.38 4.05–7.13 <.001 4.63 3.47–6.17 <.001Advanced vs early biochemical stagea 18.12 12.10–27.12 <.001 14.11 9.32–21.36 <.001Male sex 1.30 0.85–1.98 .224 – – –

UDCA treatment 0.49 0.36–0.68 <.001 – – –

Age at entry, per 10 years 1.57 1.43–1.68 <.001 1.42 1.29–1.55 <.001Year of diagnosis, per decade 0.79 0.62–0.96 .006 – – –

Bilirubin level, �ULNb 1.94 1.35–2.79 <.001 – – –

Albumin level, �LLN 0.18 0.06–0.57 <.001 – – –

ALP level, �ULNb 1.56 1.28–1.89 <.001 1.85 1.17–2.95 .009AST/ALT ratiob 1.79 1.28–2.52 <.001 – – –

Platelets, per 10-unit increase, �103/mm3 0.96 0.95–0.98 <.001 0.98 0.97–0.99 .045

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholicacid; ULN, upper normal limit of normal.aHR was obtained by considering biochemical transition as a time-dependent covariate.bThese biochemical variables were transformed with a natural logarithm.


Supplementary Table 5. Baseline Factors Associated With the Transition From Biochemically Early to Moderately AdvancedDisease in the Original Data Set

N



Male sex 1615 1.27 1.01-1.60 .043 – – –

Age at entry, per 10 years 1615 1.06 1.00-1.12 .044 – – –

Year of diagnosis, per decade 1615 0.86 0.78-0.94 <.001 – – –

UDCA use 1615 0.90 0.73-1.09 .279 0.67 0.45–1.01 .057Bilirubinb 1615 2.14 1.78-2.58 <.001 1.84 1.46–2.31 <.001Albumin 1615 0.07 0.04-0.12 <.001 0.04 0.02–0.09 <.001ALP, �ULNb 1486 1.44 1.30–1.60 <.001 1.20 1.05–1.36 .002AST, �ULNb 1406 1.44 1.26–1.65 <.001ALT, �ULN 1338 1.15 1.03–1.29 .015AST/ALT ratio 1322 1.37 1.12–1.67 .002 – – –

Platelets, per 10-unit increase 1202 0.98 0.97–0.99 <.001 0.99 0.99–0.99 .011

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholicacid; ULN, upper normal limit of normal.aThis model was constructed for 1042 patients with complete data on all variables.bThese biochemical variables were transformed with a natural logarithm.

Supplementary Table 6. The Association of Time-Dependent Transition With Clinical Events in Original Data Set

N



Moderate vs early biochemical stagea 1615 3.37 2.28–4.97 <.001 2.54 1.58–4.09 <.001Advanced vs moderate biochemical stagea 1615 5.38 4.05–7.13 <.001 4.09 2.83–5.93 <.001Advanced vs early biochemical stagea 1615 18.12 12.10–27.12 <.001 10.42 6.23–17.42 <.001Male sex 1615 1.30 0.85–1.98 .224 – – –

UDCA treatment 1615 0.49 0.36–0.68 <.001 – – –

Age at entry, per 10 years 1615 1.57 1.43–1.68 <.001 1.56 1.40–1.72 <.001Year of diagnosis, per decade 1615 0.79 0.62–0.96 .006 – – –

Bilirubin, �ULNb 1615 1.94 1.35–2.79 <.001 – – –

Albumin, �LLN 1615 0.18 0.06–0.57 <.001 – – –

ALP, �ULNc 1486 1.59 1.31–1.95 <.001 1.56 1.19–2.04 .001AST, �ULNc 1406 1.49 1.14–1.93 .003ALT, �ULNc 1338 1.06 0.84–1.33 .637AST/ALT ratioc 1322 1.92 1.30–2.82 .001 – – –

Platelets, per 10-unit increase 1202 0.96 0.94–0.98 <.001 0.97 0.95–0.99 .007

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HR, hazard ratio; LLN, lower limit of normal; UDCA, ursodeoxycholicacid; ULN, upper normal limit of normal.aThis model was constructed for 1042 patients with complete data on all variables.bHR was obtained by considering biochemical transition as a time-dependent covariate.cThese biochemical variables were transformed with a natural logarithm.


factors associated with progression and outcomes of early

Documents