Explore the vast range of titles available.

Background Support vector machines (SVM) are a powerful tool to analyze data with a number of predictors approximately equal or larger than the number of observations. However, originally, application of SVM to analyze biomedical data was limited because SVM was not designed to evaluate importance of predictor variables. Creating predictor models based on only the most relevant variables is essential in biomedical research. Currently, substantial work has been done to allow assessment of variable importance in SVM models but this work has focused on SVM implemented with linear kernels. The power of SVM as a prediction model is associated with the flexibility generated by use of non-linear kernels. Moreover, SVM has been extended to model survival outcomes. This paper extends the Recursive Feature Elimination (RFE) algorithm by proposing three approaches to rank variables based on non-linear SVM and SVM for survival analysis. Results The proposed algorithms allows visualization of each one the RFE iterations, and hence, identification of the most relevant predictors of the response variable. Using simulation studies based on time-to-event outcomes and three real datasets, we evaluate the three methods, based on pseudo-samples and kernel principal component analysis, and compare them with the original SVM-RFE algorithm for non-linear kernels. The three algorithms we proposed performed generally better than the gold standard RFE for non-linear kernels, when comparing the truly most relevant variables with the variable ranks produced by each algorithm in simulation studies. Generally, the RFE-pseudo-samples outperformed the other three methods, even when variables were assumed to be correlated in all tested scenarios. Conclusions The proposed approaches can be implemented with accuracy to select variables and assess direction and strength of associations in analysis of biomedical data using SVM for categorical or time-to-event responses. Conducting variable selection and interpreting direction and strength of associations between predictors and outcomes with the proposed approaches, particularly with the RFE-pseudo-samples approach can be implemented with accuracy when analyzing biomedical data. These approaches, perform better than the classical RFE of Guyon for realistic scenarios about the structure of biomedical data.

Primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) are reported to have the best outcomes after liver transplantation. Based on excellent 5-yr survival results after transplantation, it has been suggested that PSC patients may benefit from \"preemptive\" transplantation to reduce the risk of cholangiocarcinoma. In this study, we compared 10-yr survival of patients with PSC and PBC using a large database after adjusting for other confounding risk factors. The United Network for Organ Sharing (UNOS) database of all patients who had liver transplantation from 1987 to 2001 was used for analysis after excluding patients with multiple organ transplantation, children, and incomplete data. Patients with PSC (n = 3,309) were younger than those with PBC (n = 3,254). Retransplantation rate was high in PSC (12.4%vs 8.5%; p< 0.01), and PSC was an independent predictor for retransplantation on multivariate analysis. Cox regression analysis showed that PSC patients had significantly lower graft and patient survival compared to PBC patients after adjusting for other risk factors. Lower survival in PSC became apparent 7 yr after transplantation. Patients with PSC had a higher retransplantation rate and lower survival when compared to PBC. Based on this analysis, we do not recommend preemptive liver transplantation for patients with PSC.

Patients with cirrhosis and men have been under-represented in most studies examining the clinical benefit of response to ursodeoxycholic acid (UDCA) in primary biliary cholangitis (PBC). The aim of this study was to study the association of UDCA response and liver-related death or transplantation, hepatic decompensation, and hepatocellular carcinoma (HCC) in patients with PBC cirrhosis. We conducted a retrospective cohort study of veterans, predominantly men, with PBC and compensated cirrhosis to assess the association of UDCA response with the development of all-cause and liver-related mortality or transplantation, hepatic decompensation, and HCC using competing risk time-updating Cox proportional hazards models. We identified 501 subjects with PBC and compensated cirrhosis, including 287 UDCA responders (1,692.8 patient-years [PY] of follow-up) and 214 partial responders (838.9 PY of follow-up). The unadjusted rates of hepatic decompensation (3.8 vs 7.9 per 100 PY, P < 0.0001) and liver-related death or transplantation (3.7 vs 6.2 per 100 PY, P < 0.0001) were lower in UDCA responders compared with partial responders. UDCA response was associated with a lower risk of hepatic decompensation (subhazard ratio [sHR] 0.54, 95% confidence interval [CI] 0.31-0.95, P = 0.03), death from any cause or transplantation (adjusted hazard ratio 0.49, 95% CI 0.33-0.72, P = 0.0002), and liver-related death or transplantation (sHR 0.40, 95% CI 0.24-0.67, P = 0.0004), but not HCC (sHR 0.39, 95% CI 0.60-2.55, P = 0.32). In a sensitivity analysis, the presence of portal hypertension was associated with the highest UDCA-associated effect. UDCA response is associated with a reduction in decompensation, all-cause, and liver-related death or transplantation in a cohort of predominantly male patients with cirrhosis, with the highest benefit in patients with portal hypertension.

Magnetic resonance (MR) risk scores and liver stiffness (LS) have individually been shown to predict clinical outcomes in primary sclerosing cholangitis (PSC). The aim of this study was to assess their complementary prognostic value. Patients with PSC from 3 European centers with a 3-dimensional MR cholangiography available for central reviewing and a valid LS measurement assessed by vibration-controlled transient elastography by FibroScan performed within a 6-month interval were included in a longitudinal retrospective study. The MR score (Anali) without gadolinium (Gd) was calculated according to the formula: (1 × dilatation of intrahepatic bile ducts) + (2 × dysmorphy) + (1 × portal hypertension). The primary end point was survival without liver transplantation or cirrhosis decompensation. The prognostic values of LS and Anali score without Gd were assessed using Cox proportional hazard models. One hundred sixty-two patients were included. Over a total follow-up of 753 patient-years, 40 patients experienced an adverse outcome (4 liver transplantations, 6 liver-related deaths, and 30 cirrhosis decompensations). LS and Anali score without Gd were significantly correlated (ρ = 0.51, P < 0.001) and were independently associated with the occurrence of an adverse outcome. Optimal prognostic thresholds were 10.5 kPa for LS and 2 for the Anali score without Gd. Hazard ratios (95% confidence interval) were 2.07 (1.06-4.06) and 3.78 (1.67-8.59), respectively. The use in combination of these 2 thresholds allowed us to separate patients into low-, medium-, and high-risk groups for developing adverse outcomes. The 5-year cumulative rates of adverse outcome in these 3 groups were 8%, 16%, and 38% (P < 0.001), respectively. The combined use of MRI and vibration-controlled transient elastography permits easy risk stratification of patients with PSC.

Risk stratification based on biochemical variables is a useful tool for monitoring ursodeoxycholic acid (UDCA)-treated patients with primary biliary cholangitis (PBC). Several UDCA response criteria and scoring systems have been proposed for risk prediction in PBC, but these have not been validated in large external cohorts. We performed a study on data of 1746 UDCA-treated patients with PBC from 25 centers in Europe, United States, and Canada. The prognostic performance of the risk scoring systems (GLOBE and UK-PBC) and the UDCA response criteria (Barcelona, Paris I, Paris II, Rotterdam, and Toronto) were evaluated. We regarded cirrhosis-related complications (ascites, variceal bleeding, and/or hepatic encephalopathy) as clinical end points. A total of 171 patients reached a clinical end point during a median 7 years (range 1-16 years) of follow-up. The 5-, 10- and 15-year adverse outcome-free survivals were 95%, 85%, and 77%. The GLOBE and UK-PBC scores predicted cirrhosis-related complications better than the UDCA response criteria. The hazard ratio (HR) for a 1 standard deviation increase was HR 5.05 (95% confidence interval (CI): 4.43-5.74, P < 0.001) for the GLOBE score and HR 3.39 (95% CI: 3.10-3.72, P < 0.001) for the UK-PBC score. Overall, the GLOBE and UK-PBC risk scores showed similar and excellent prognostic performance (C-statistic, 0.93; 95% CI: 0.91%-95% vs 0.94; 95% CI: 0.91%-0.96%). In our international, multicenter PBC cohort, the GLOBE and UK-PBC risk scoring systems were good predictors of future cirrhosis-related complications.

There is a paucity of information on extrahepatic autoimmune (EHA) conditions associated with primary biliary cirrhosis (PBC) and on the impact of EHA conditions on PBC patients’ survival. Our goal was to assess the association between PBC and other autoimmune diseases and the impact of EHA conditions on the natural history of PBC. We took advantage of 361 consecutive PBC patients enrolled between 1975 and 2012 (22 males, 339 females; mean follow-up 8 ± 6.9 years). Any associated EHA conditions, PBC histological stage at diagnosis, biochemical data, physiological history, and extrahepatic malignancies developing during the follow-up were recorded. Survival was analyzed by means of Kaplan-Meier curves. Importantly, 221 patients (61.2 %) had at least one EHA conditions: 45 patients (20.4 %) had Hashimoto thyroiditis; 7 (3.2 %) had Graves’ thyroiditis; 65 (29.4 %) had Raynaud’s phenomenon; 124 (56.1 %) had Sjogren’s syndrome; 8 (3.6 %) had systemic lupus erythematosus; 22 (9.9 %) had scleroderma; 22 (9.9 %) had rheumatoid arthritis; 18 (8.1 %) had cutaneous autoimmune diseases; 8 (3.6 %) had vasculitis; 5 (1.4 %) had celiac disease; and 25 (13.1 %) had other EHA conditions. The proportion of patients with associated EHA conditions enrolled during representative periods (1975–1980, 1981–1990, 1991–2000, 2001–2010, 2011–2012) remained stable. No differences emerged between patients with versus without EHA conditions in terms of mean age at PBC diagnosis, antimitochondrial antibody (AMA), or antinuclear antibody (ANA) positivity, histological stage at diagnosis, smoking habits, alcohol consumption, or BMI >25. Multiple logistic regression analysis showed that only female gender was significantly associated with positivity for EHA conditions (OR 4.8; 95 % CI 1.6–13.7, p  = 0.004). The mean survival after the diagnosis of PBC was much the same in patients with and without EHA conditions. In conclusion, EHA conditions are often associated with PBC, especially in female patients, but they do not reduce patient survival.

Ductopenia drives biochemical failure and histological progression in primary biliary cholangitis (PBC), influencing its course and prognosis, but its prevalence, features, and prognosis remain unclear. This study aimed to characterize ductopenia in PBC and identify early predictive biomarkers. From August 2013 to April 2025, 518 of the biopsy-proven PBC patients were enrolled, analyzed for demographics, pathology, and clinical features, and grouped by ductopenia presence. 201 patients were followed until June 15, 2025, with liver-related adverse events (including TIPS, splenectomy with portosystemic shunt or portoazygous devascularization, liver failure, death, or liver transplantation) as endpoints. Kaplan-Meier and Cox regression assessed prognosis. The overall proportion of patients with PBC and ductopenia was 56.76% (294/518), Notably, ductopenia was present in 24.83% (74/298) of patients with early-stage disease. Compared with the group without ductopenia, the ductopenia group showed significantly higher levels of cholestasis indicators (such as TBIL, ALP, GGT, and TBA) and autoantibodies (ANA, AMA anti-gp210), but significantly lower levels of liver synthetic function indicators (such as ALB and cholinesterase) and blood components (RBC, PLT, and HGB) (all <0.05). The median follow-up time was 7.60 years (interquartile range: 5.80-9.20 years). The prevalence of liver-related adverse events was significantly higher in PBC patients with ductopenia than in those without ( <0.05). Cox regression analysis confirmed that ductopenia (HR=8.868, 95% CI: 1.135-69.307, =0.037) was an independent risk factor for the occurrence of liver-related adverse events in patients with PBC. Multivariable logistic regression analysis identified that serum ANA(≥1:1000) (OR= 2.180, 95% CI: 1.261-3.769), elevated GGT (OR = 1.002, 95% CI: 1.001-1.003, = 0.001) and TBIL (OR= 1.020, 95% CI: 1.005-1.035), lowed ALB (OR= 0.943, 95% CI: 0.896-0.993) as biomarkers for ductopenia in patients with early-stage PBC. Ductopenia is relatively common in patients with PBC, and its prevalence significantly increases with disease progression. Ductopenia was an independent risk factor for the occurrence of liver-related adverse events in patients with PBC. ANA(≥1:1000), TBIL, GGT, and ALB are early predictive biomarkers for ductopenia in patients with PBC.

We used data from the Fibrotic Liver Disease Consortium to evaluate the impact of ursodeoxycholic acid (UDCA) treatment across race/ethnicity, gender, and clinical status among patients with primary biliary cholangitis. Data were collected from \"index date\" (baseline) through December 31, 2016. Inverse Probability of Treatment Weighting was used to adjust for UDCA treatment selection bias. Cox regression, focusing on UDCA-by-risk factor interactions, was used to assess the association between treatment and mortality and liver transplant/death. Among 4,238 patients with primary biliary cholangitis (13% men; 8% African American, 7% Asian American/American Indian/Pacific Island [ASINPI]; 21% Hispanic), 78% had ever received UDCA. The final multivariable model for mortality retained age, household income, comorbidity score, total bilirubin, albumin, alkaline phosphatase, and interactions of UDCA with race, gender, and aspartate aminotransferase/alanine aminotransferase ≥1.1. Among untreated patients, African Americans and ASINPIs had higher mortality than whites (adjusted hazard ratio [aHR] = 1.34, 95% confidence interval [CI] 1.08-1.67 and aHR = 1.40, 95% CI 1.11-1.76, respectively). Among treated patients, this relationship was reversed (aHR = 0.67, 95% CI 0.51-0.86 and aHR = 0.88, 95% CI 0.67-1.16). Patterns were similar for liver transplant/death. UDCA reduced the risk of liver transplant/death in all patient groups and mortality across all groups except white women with aspartate aminotransferase/alanine aminotransferase ≥1.1. As compared to patients with low-normal bilirubin at baseline (≤0.4 mg/dL), those with high-normal (1.0 > 0.7) and mid-normal bilirubin (0.7 > 0.4) had significantly higher liver transplant/death and all-cause mortality. African American and ASINPI patients who did not receive UDCA had significantly higher mortality than white patients. Among African Americans, treatment was associated with significantly lower mortality. Regardless of UDCA treatment, higher baseline bilirubin, even within the normal range, was associated with increased mortality and liver transplant/death compared with low-normal levels.

Although a variant of primary biliary cirrhosis (PBC) characterized by features of autoimmune hepatitis (AIH) has been recognized for many years, few studies with ample numbers of patients have focused on its natural history. This study aimed to clarify the natural history, prognosis, and response to therapy in a cohort of patients with PBC with AIH features. We retrospectively analyzed 277 PBC patients without AIH features and 46 PBC patients with AIH features seen between September 2004 and April 2014. The 5-year adverse outcome-free survival of PBC patients with AIH features was 58 % compared to 81 % in PBC patients without AIH features. Multivariate analysis in the patients with AIH features indicated that total bilirubin ≥2.70× the upper limit of normal predicted a poor prognosis ( p  = 0.008, relative risk 8.39, 95% confidence interval (CI) 1.73, 40.73). Combination therapy with ursodeoxycholic acid (UDCA) and immunosuppression provided better short-term responses in PBC patients with AIH features, defined by multiple criteria. Higher aspartate aminotransferase (AST) level at accession suggested better prognosis for PBC patients with AIH features while worse prognosis for PBC patients without AIH features. PBC patients with AIH features differ from those without AIH features in terms of natural history, prognostic indicators, and response to therapy.