Carbone, Marco

AbstractBackground & AimsMachine learning (ML) provides new approaches for prognostication through the identification of novel subgroups of patients. We explored whether ML could support disease sub‐phenotyping and risk stratification in primary biliary cholangitis (PBC).MethodsML was applied to an international dataset of PBC patients. The dataset was split into a derivation cohort (training set) and a validation cohort (validation set), and key clinical features were analysed. The outcome was a composite of liver‐related death or liver transplantation. ML and standard survival analysis were performed.ResultsThe training set was composed of 11,819 subjects, while the validation set was composed of 1,069 subjects. ML identified four clusters of patients characterized by different phenotypes and long‐term prognosis. Cluster 1 (n = 3566) included patients with excellent prognosis, whereas Cluster 2 (n = 3966) consisted of individuals at worse prognosis differing from Cluster 1 only for albumin levels around the limit of normal. Cluster 3 (n = 2379) included young patients with florid cholestasis and Cluster 4 (n = 1908) comprised advanced cases. Further sub‐analyses on the dynamics of albumin within the normal range revealed that ursodeoxycholic acid‐induced increase of albumin >1.2 x lower limit of normal (LLN) is associated with improved transplant‐free survival.ConclusionsUnsupervised ML identified four novel groups of PBC patients with different phenotypes and prognosis and highlighted subtle variations of albumin within the normal range. Therapy‐induced increase of albumin >1.2 x LLN should be considered a treatment goal.

Changes over time in the presenting features and clinical course of patients with primary biliary cholangitis are poorly described. We sought to describe temporal trends in patient and disease characteristics over a 44‐year period across a large international primary biliary cholangitis cohort of 4,805 patients diagnosed between 1970 and 2014, from 17 centers across Europe and North America. Patients were divided into five cohorts according to their year of diagnosis: 1970‐1979 (n = 143), 1980‐1989 (n = 858), 1990‐1999 (n = 1,754), 2000‐2009 (n = 1,815), and ≥2010 (n = 235). Age at diagnosis, disease stage, response to ursodeoxycholic acid, and clinical outcomes were compared. Mean age at diagnosis increased incrementally by 2‐3 years per decade from 46.9 ± 10.1 years in the 1970s to 57.0 ± 12.1 years from 2010 onward (P < 0.001). The female to male ratio (9:1) and antimitochondrial antibody positivity (90%) were not significantly variable. The proportion of patients presenting with mild biochemical disease (according to Rotterdam staging) increased from 41.3% in the 1970s to 72.2% in the 1990s (P < 0.001) and remained relatively stable thereafter. Patients with a mild histological stage at diagnosis increased from 60.4% (1970‐1989) to 76.5% (1990‐2014) (P < 0.001). Correspondingly, response to ursodeoxycholic acid according to Paris‐I criteria increased; 51.7% in the 1970s and 70.5% in the 1990s (P < 0.001). Recent decades were also characterized by lower decompensation rates (18.5% in the 1970s to 5.8% in the 2000s, P < 0.001) and higher 10‐year transplant‐free survival (48.4%, 68.7%, 79.7%, and 80.1% for each respective cohort; P < 0.001). Conclusion: In recent decades, a pattern of primary biliary cholangitis presentation consistent with an older age at diagnosis alongside reduced disease severity has been noted; the observed trends may be explained by an increase in routine testing of liver function and/or a changing environmental trigger. (Hepatology 2018;67:1920‐1930).

Primary sclerosing cholangitis (PSC) is strongly associated with several human leukocyte antigen (HLA) haplotypes. Due to extensive linkage disequilibrium and multiple polymorphic candidate genes in the HLA complex, identifying the alleles responsible for these associations has proven difficult. We aimed to evaluate whether studying populations of admixed or non‐European descent could help in defining the causative HLA alleles. When assessing haplotypes carrying HLA‐DRB1*13:01 (hypothesized to specifically increase the susceptibility to chronic cholangitis), we observed that every haplotype in the Scandinavian PSC population carried HLA‐DQB1*06:03. In contrast, only 65% of HLA‐DRB1*13:01 haplotypes in an admixed/non‐European PSC population carried this allele, suggesting that further assessments of the PSC‐associated haplotype HLA‐DRB1*13:01‐DQA1*01:03‐DQB1*06:03 in admixed or multi‐ethnic populations could aid in identifying the causative allele.