Study on Clinical and Pathological Features of Primary Biliary Cholangitis and Risk Factors Related to Disease Progression

The study subjects were derived from all hospitalized and outpatient patients who attended the Second Department of Hepatology at Beijing Ditan Hospital, Capital Medical University, between January 1, 2015, and June 31, 2026, and who underwent liver biopsy with a confirmed diagnosis of PBC; all patients were from Beijing Ditan Hospital, Capital Medical University, and case screening and data collection were performed through the hospital's HIS and LIS systems. The inclusion criteria were as follows: (1) met the diagnostic criteria for PBC according to the Guidelines for the Diagnosis and Treatment of Primary Biliary Cholangitis (2021) and the Guidelines for the Diagnosis and Treatment of Primary Biliary Cholangitis (2025 Edition), satisfying at least two of three criteria-① biochemical evidence of cholestasis (elevated ALP and/or GGT) with imaging excluding extrahepatic or intrahepatic large bile duct obstruction, ② positive AMAs/AMA-M2 or positivity for other PBC-specific autoantibodies (anti-gp210, anti-sp100), and ③ histologic evidence of non-suppurative destructive cholangitis and small bile duct destruction; (2) underwent liver biopsy with a complete pathological report; and (3) had complete clinical data with missing values for key variables not exceeding 20%. The exclusion criteria comprised: (1) concomitant other liver diseases such as chronic hepatitis B, C, D, or E, alcoholic liver disease, non-alcoholic fatty liver disease, drug-induced liver injury, autoimmune hepatitis, primary hemochromatosis, or Wilson's disease; (2) concurrent non-hepatotropic viral infections causing liver injury, including Epstein-Barr virus, cytomegalovirus, and human immunodeficiency virus; (3) concurrent liver malignancy; (4) age under 18 years; and (5) severely incomplete clinical data with missing values for key variables exceeding 20%. This was a retrospective study based on all consecutive cases meeting the inclusion and exclusion criteria at Beijing Ditan Hospital from January 2015 to June 2026; based on preliminary study data and published literature, approximately 300-500 PBC patients were expected to be enrolled. Sample size estimation was based on the following rationale: according to previous studies, the incidence of disease progression (decompensated cirrhosis, liver transplantation, or death) among PBC patients was approximately 15%-30%; with α = 0.05 (two-sided), power 1-β = 0.80, an anticipated 10-15 independent variables, and following the rule of events per variable ≥ 10, at least 100-150 progression events were required, and this study was expected to enroll 300-500 patients with an estimated 60-150 progression events, which would meet the sample size requirements for multivariate regression analysis. The following patient data were retrospectively collected through the hospital HIS and LIS systems, and all tests and examinations were performed by the Clinical Laboratory, Radiology Imaging Center, Department of Pathology, and Gastroscopy Center of Beijing Ditan Hospital. The variables collected included: (1) general demographic data (sex, age, age at diagnosis, disease duration in months, and family history of PBC or autoimmune diseases); (2) clinical data, comprising clinical symptoms (fatigue, pruritus, abdominal distension, poor appetite, jaundice, xerostomia and xerophthalmia, recorded as present/absent based on medical records), comorbidities (extrahepatic autoimmune diseases including Sjögren's syndrome, thyroiditis, CREST syndrome, and rheumatoid arthritis, as well as hypertension, diabetes mellitus, and hyperlipidemia), smoking history, alcohol consumption history, and history of drug allergies; (3) laboratory parameters obtained within one week before liver biopsy, including complete blood count (WBC, RBC, HGB, PLT), liver function tests (ALT, AST, TBIL, DBIL, ALB, GLB, A/G ratio, GGT, ALP, CHE, TBA), coagulation function (PT, PTA, INR), lipid profile (TC, TG, HDL-C, LDL-C), renal function (Cr, UA, eGFR), immunological parameters (IgG, IgM), and autoantibodies (AMAs/AMA-M2, ANA and its patterns, ACA, anti-sp100, anti-gp210, anti-Ro-52); (4) UDCA treatment and response assessment, including treatment status (yes/no), dose in mg/kg/day, time of initiation, and biochemical response after one year according to the Paris, Barcelona, or Toronto criteria; (5) imaging and non-invasive examinations within one month before biopsy and during follow-up (with follow-up data retrospectively collected from clinical records), comprising abdominal ultrasonography (spleen thickness and length, main portal vein width, ascites), gastroscopy (esophageal and gastric varices, present/absent and grade), and transient elastography (liver stiffness measurement in kPa); (6) pathological data, including histological stage according to the Ludwig system (stages I-IV) and pathological features such as florid duct lesion, bile duct loss/ductopenia (present/absent and degree), perisinusoidal fibrosis (present/absent and degree), copper-associated protein deposition (present/absent and degree), interface hepatitis, cholestasis, and lymphoid follicle aggregation (all recorded as present/absent); and (7) follow-up and outcome data, comprising follow-up duration from diagnosis to last follow-up or endpoint event, endpoint events defined as disease progression (histological stage progression, decompensated cirrhosis events including ascites, variceal bleeding, and hepatic encephalopathy, liver transplantation, or liver-related death), and ALP and TBIL levels at last follow-up with UDCA response status. A study database was established using EpiData or Excel, with data entered independently by two individuals and the database locked after verification; data were stored on an encrypted hospital intranet server accessible only to project team members. This was a retrospective study and did not involve biological sample collection, nor did it involve new laboratory testing; all laboratory parameters were derived from routine clinical testing at Beijing Ditan Hospital, with complete blood count measured using an automated hematology analyzer, liver function, renal function, and lipid profile measured using an automated biochemical analyzer, coagulation function measured using an automated coagulation analyzer, immunoglobulins measured using immunoturbidimetry, and autoantibodies detected using indirect immunofluorescence for ANA and immunoblotting for AMAs, anti-gp210, anti-sp100, ACA, and anti-Ro-52. For statistical analysis, SPSS 27.0 and R 4.2 were used. Normally distributed or approximately normally distributed continuous variables were expressed as mean ± standard deviation, severely skewed continuous variables as median (interquartile range), and categorical variables as counts (percentages). For between-group comparisons, normally distributed continuous variables were compared between two groups using the independent-samples t-test and among multiple groups using one-way ANOVA; skewed continuous variables were compared between two groups using the Mann-Whitney U test and among multiple groups using the Kruskal-Wallis H test; categorical variables were compared using the chi-square test or Fisher's exact test when expected frequencies were less than 5; all tests were two-sided and P < 0.05 was considered statistically significant. Kaplan-Meier curves were used to estimate event-free survival rates across groups, with between-group differences compared using the log-rank test. Univariate and multivariate Cox proportional hazards regression models were used to analyze independent risk factors for disease progression, with hazard ratios and 95% confidence intervals calculated. Variables with P < 0.10 in univariate analysis were entered into the multivariate Cox model, and independent risk factors were selected using the forward LR method. GLOBE and UK-PBC risk scores were calculated, and time-dependent ROC curves were used to evaluate the predictive performance of each score and of the model established in this study, with the area under the curve calculated. Potential confounders (age, sex, disease duration) were adjusted for by including them as covariates in the Cox model. For missing data, key variables with a missing rate below 5% were imputed using the median (continuous) or mode (categorical), those with 5%-20% missing were handled by multiple imputation, and variables with over 20% missing were excluded from multivariate analysis. Prespecified subgroup analyses were performed by histological stage (I/II vs III/IV), by UDCA response (responders vs non-responders), and by autoantibody type (anti-gp210 positive vs negative). To test robustness, sensitivity analyses were performed using different definitions of disease progression, such as using only decompensated cirrhosis or liver-related death as the endpoint.