The intestinal mucosa is host to a large number of commensal microbes, which in humans are dominated by the bacterial phyla Firmicutes and Bacteroides and less so Actinobacteria and Proteobacteria. Commensal flora resident in the gastrointestinal tract have co-evolved with man and, through a symbiotic relationship, perform a broad range of essential physiological functions ranging from the promotion of intestinal defences against outgrowth of pathogenic species to a hitherto unknown level of host-related co-metabolism [7]. The intestinal epithelial monolayer underneath provides a structural barrier that prevents invasion of bacteria beyond the gut and also expresses a variety of PRR, such as toll-like receptors (TLR) and nucleotide-binding oligomerisation domain-containing proteins (NOD)-1 and NOD2, which are involved in the recognition of cellular injury and damage [8]. The ability of the intestinal epithelial cells to secrete cytokines and chemokines in response to enteric microbial fluctuations, pathogens or injury allows them to actively shape local immune responses and modulate sub-epithelial dendritic cell (DC) and lymphocyte positioning and activation. A degree of mutualism between commensal flora and the human host is illustrated by the fact that the activation of epithelial TLR2 or TLR9 increases gut barrier function, whereas mice deficient in the downstream TLR-signalling molecule MyD88 are susceptible to experimentally induced IBD [9]. However, it remains tentative how exactly the mucosal epithelium distinguishes between commensal bacteria and pathogenic strains. Disturbances in immune or epithelial homeostasis can lead to gut inflammation, and in certain circumstances, commensal flora may act as pathogens. Evidence to support impaired tolerance to commensal flora in the pathogenesis of IBD is provided by the interleukin (Il)-10 ^−/− murine model, wherein intestinal inflammation is abrogated under germ-free conditions and in Il-2-deficient animals which develop colitis following enteric exposure to Escherichia coli but not Bacteroides vulgates [10, 11].

Ordinarily, gut commensals and pathogens are confined to the gut by the mucosal epithelium and mesenteric lymph nodes (MLN). It is conceivable, however, that in the presence of intestinal inflammation and a disturbed epithelial barrier, bacteria can enter the portal circulation and the liver where further levels of regulation exist to prevent uncontrolled systemic immune activation. Under normal circumstances, liver-resident antigen-presenting cells (APC) display attenuated responses to endotoxin exposure, and the liver functions as a second ‘firewall’ that clears commensals from the circulation if intestinal defences are overwhelmed [12]. Intestinal CX₃CR₁ ⁺ macrophages are a critical component of the intestinal barrier and express TLR to sense microorganisms and activate innate lymphoid cells (ILC) to secrete IL-22, thereby directly promoting epithelial integrity and repair [13]. Deletion of CX₃CR₁ in mice not only results in increased bacterial translocation to MLN and susceptibility to colitis, but in a diet-induced model of liver disease to hepatitis, demonstrating how defects in gut integrity can drive hepatic inflammation [14]. Compositional changes to the gut microbiota as a consequence of defective inflammasome pathway signalling have also been shown to induce liver inflammation driven by hepatic TNFα activation as a consequence of TLR4 and TLR9 agonists in the portal circulation [15]. Such observations indicate how defective pathogen sensing as a consequence of a divergent enteric microbiome, or perhaps through genetic variations affecting the threshold for PRR signalling [16], might change the gut microbiota leading to hepatic inflammation in association with IBD. This concept is supported by data from mouse models in which changes in intestinal bacterial populations or infusion of bacterial antigens into the portal circulation lead to peri-cholangitis [17].

Analysis of colonic mucosal biopsies in British patients with PSC and IBD suggests a restriction in the biodiversity of adherent microbiota, highlighted by a near absence of Bacteroides and by significant increases in Escherichia, Lachnospiraceae and Megasphaera, when compared to individuals with IBD alone and healthy controls [5, 18]. Restricted biodiversity was also reported when analysing stool samples in a Norwegian PSC cohort, albeit with a marked increase in the anaerobic genus Veillonella [19]. These subtle, yet important, differences are likely to reflect the impact of dietary and environmental influences on enteric flora composition and possibly disparate genetic risk factors that alter microbial handling between patient populations [20, 21].

Seropositivity for perinuclear anti-neutrophil cytoplasmic antibodies (pANCA) is frequently observed in patients with PSC, and although not disease specific, these antibodies are highly reactive against the microbial cell division protein FtsZ present in virtually every bacteria of the intestinal microflora [22]. Some autoantibodies detected in PSC also bind to biliary epithelium, which, as a contiguous component of the mucosal epithelium, can be activated through enhanced TLR4 and TLR9 expression resulting in the secretion of pro-inflammatory cytokines and chemokines that drive hepatobiliary inflammation [23].

Based on pathological observations, hepatic innate immune responses have traditionally been considered a primary inciting event in PSC, with disease development initiated via bacteria or pathogen-associated molecular patterns (PAMPs) that enter the portal circulation via a permeable intestinal mucosa [24]. Progressive injury occurs in small, medium and large bile ducts with resultant inflammation and concentric periductal fibrosis. In early disease, these changes are confined to portal tracts, with a mixed inflammatory cell infiltrate composed of natural killer (NK)-cells, NKT-cells, B-cells, macrophages and predominantly activated effector or memory T-cells [25]. The ratio of CD4⁺ to CD8⁺ T-cells in PSC patients shows considerable inconsistencies in different studies although CD4⁺ T-cells are more commonly observed in the portal tracts, whereas CD8⁺ T-cells populate portal areas and the parenchyma [26]. DCs which prime intrahepatic T-cell responses are inherently tolerogenic in normal liver and express low levels of the co-stimulatory molecules required for full T-cell activation [27]. Furthermore, liver-derived DCs preferentially secrete the immunoregulatory cytokine IL-10 and are also capable of inducing peripheral regulatory T-cells (T_reg), a cell population critical for suppression of immune responses. However, patients with PSC often exhibit reduced T-cell expression of the IL-2 receptor, particularly in association with IL-2Rα genetic polymorphisms and defective T_reg responses [28]. These observations are of clinical interest given that Il2Rα ^−/− mice develop spontaneous T-cell-mediated cholangitis and colitis in the context of defective regulatory T_reg activity [29].

A subpopulation of CD4⁺ and CD8⁺ T-cells have been identified, which lack the co-stimulatory molecule CD28. These CD28-negative T-cells are IL2Rα^lo and enriched in PSC relative to normal liver, where they express phenotypic surface markers in keeping with an activated memory phenotype (CD45RA⁻ CCR7⁻) in addition to containing high quantities of cytotoxic molecules granzyme B and perforin. Liver-infiltrating CD28⁻ T-cells are equipped with chemokine homing receptors that facilitate recruitment and positioning in proximity to bile ducts, where they induce activation of biliary epithelial cells (BEC) [26]. CD28-negative T-cells appear to be chronically activated immunopathogenic lymphocytes that are less susceptible to regulation by conventional T_reg, making them potentially important drivers of hepatobiliary inflammation.

Defective numbers of functional intrahepatic and peripheral blood T_reg are implicated across the spectrum of autoimmunity, often within the expanse of heightened effector T_h17 responses to pathogen stimulation [30]. IL-17-producing cells, including mucosal-associated invariant T-cells (MAIT-cells), T_c17-cells and T_h17-cells, are abundant in the intestinal lamina propria as well as the liver. In the gut, they are maintained by commensal bacteria which induce innate lymphoid cells (ILC) to secrete IL-22, a critical factor for IL-17A expression in T-cells [31].

Intra-organ recruitment and localisation of leucocytes is a highly coordinated process regulated through the selective expression of integrins and chemokine receptors (expressed by lymphocytes), which allow interactions with tissue-specific adhesion molecules and chemokine ligands expressed by endothelial structures – the first port of entry into solid organs within the human body. In the gut, the endothelial phenotype is highlighted by expression of the chemokine CCL25 and the adhesion molecule mucosal addressin cell adhesion molecule-1 (MAdCAM)-1. This pairing facilitates recruitment of lymphocytes imprinted with gut tropism; specifically those which express the chemokine receptor CCR9 and integrin α4β7. Under homeostatic conditions, MAdCAM-1 and CCL25 are critically involved in the recruitment of lymphocytes to the lamina propria; and although their expression increases significantly during inflammation [32], effector cells may also engage other adhesion molecules including vascular adhesion protein (VAP)-1, intercellular adhesion molecule (ICAM)-1 and vascular cell adhesion molecule (VCAM)-1, which become induced in response to pro-inflammatory signals. In addition, human lamina propria lymphocytes express chemokine receptors CCR2, CCR5, CCR6 and CXCR3, which might play a more pertinent role than CCR9 during later stages of intestinal inflammation [33].

Lymphocytes can enter the liver via several routes, including post-capillary portal venules and hepatic sinusoids. Effector T-cells infiltrating the inflamed human liver commonly express high levels of CXCR3, which allow them to respond to the chemokine ligands CXCL9 and CXCL10, both of which are induced by interferon (IFN)-γ and tumour necrosis factor (TNF)α and detectable on sinusoidal endothelium [34]. CXCR3 appears to be a common signal for recruitment of T_h1, T_h17 and T_reg, although subsequent positioning within the liver may involve other chemokines including CCR4 and CCR10, which localise T_reg in portal tracts. The chemokine receptor CXCR6 may have a role in recruitment of effector T-cells and the patrolling of sinusoids by NKT-cells [34].

Animal models of immune-mediated hepatobiliary injury have demonstrated inflammation-dependent roles for ICAM-1, VCAM-1 and several pro-inflammatory chemokines, although no liver specific adhesion molecules have been identified. The liver endothelial phenotype is also characterised by expression of adhesion molecules more commonly associated with lymphatic structures such as VAP-1 and common lymphatic endothelial and vascular endothelial receptor-1 (CLEVER-1) [6]. Moreover, in PSC, MAdCAM-1 and CCL25 are also detectable on liver endothelium, and collectively facilitate the recruitment of α4β7⁺CCR9⁺ mucosal lymphocytes directly from the gut [35, 36]. Most of the liver-infiltrating α4β7⁺CCR9⁺ cells in PSC are CD45RA⁻ CCR7⁻ CD11a^hi and secrete interferon (IFN)-γ upon stimulation in vitro, in keeping with a long-lived memory phenotype. This suggests that upon entry to the liver, CCR9⁺ T-cells can become reactivated following exposure to a common antigenic trigger shared with the intestine [35].

The exact factors leading to aberrant expression of MAdCAM-1 in the PSC liver are incompletely understood. Recent work has demonstrated that catabolism of biological amines by VAP-1, an adhesion molecule upregulated in the inflamed liver which also behaves as a selective amine oxidase, is able to induce expression of functional MAdCAM-1 on the surface of hepatic sinusoidal endothelial cells (HSEC) [37]. Increased levels of enterobacterial-derived amines, perhaps due to enhanced absorption via the inflamed gut, may thus act as a substrate for VAP-1, thereby increasing MAdCAM-1 expression in the inflamed mucosa and hepatic endothelium. Such mechanisms could promote the unrestrained recruitment of mucosal effector cells and result in tissue damage that is characteristic of both IBD and PSC.

The capacity to imprint gut tropism on naïve lymphocytes is dependent on the ability of APC to convert retinol to all-trans retinoic acid (ATRA), which complexes with intracellular retinoid receptors to activate transcription of gut-homing receptors (CCR9 and α4β7). This ability can be conferred in vitro by the isolated addition of exogenous ATRA suggesting that retinoids in bile or stored in stellate cells, one of the main storage cells for RA, might be able to provide ATRA within the liver, allowing liver APCs to induce a gut-homing phenotype. However, isolated human hepatic DC and stellate cells are unable to induce high-level α4β7 or CCR9 expression on T-cells suggesting that the ability to imprint naïve lymphocytes with gut tropism is restricted to intestinal DCs [38]. Nevertheless, a study in mice by Neumann et al. illustrated that priming of naïve T-cells by HSEC is able to induce expression of α4β7 on CD4⁺ T-cells in a retinoic acid dependent manner [39]. Of interest, there is emerging evidence from spontaneous models of ileitis, which show that the onset of portal and lobular liver inflammation may actually develop before that observed in the gut [40].

Gut-derived lymphocytes also utilise other chemokine receptors utilise to localise to biliary epithelium, following which they can destroy bile ducts. Moreover, the biliary and intestinal epithelia are in direct mucosal continuity with one another, thus it is perhaps unsurprising that these two structures share certain chemokine expression patterns. For instance, some of the α4β7⁺ T-cells also express αEβ7⁺ which binds to E-cadherin expressed on the adherens junction of gut epithelial cells or bile ducts [36]. Similarly, the inflamed biliary epithelium expresses CCL20, and effector IL-17-secreting T-cells that infiltrate the human liver express high levels of its cognate receptor CCR6, facilitating their recruitment and positioning around bile ducts [41].

The highly polymorphic major histocompatibility complex (MHC) has been implicated in the aetiopathogenesis of human autoimmunity for decades [59], with strong albeit distinct HLA signals recently confirmed for autoimmune liver disease through GWAS. Comprehension of how HLA impacts cholestatic disease mechanistically is somewhat limited, although the fact that an association has been identified in the first instance suggests a defect in the direction and precision of T-cell-related and antigen-specific immune responses. Variation within the MHC region represents the most significant genetic risk factor for PSC [60], with proposed single-nucleotide polymorphisms in near-perfect linkage disequilibrium with alleles at both HLA-B (HLA class I region) and DRB1 (HLA class II region). Despite the striking coexistence with colitis, the majority of HLA associations in PSC are distinct from those identified in IBD [56, 61]. A key opportunity for post-GWAS research is the delineation of the antigenic repertoire of PSC-associated HLA types and potential specific triggers of T-cell activation in PSC.

Genetic links to mucosal immunity are particularly evident in PSC [6]. The importance of IL-2/IL-2Rα polymorphisms, suggested through associations at the 4q27 and 10p15 loci, respectively [56], is supported by the fact that mice lacking IL-2Rα develop autoantibodies and a T-cell-mediated cholangitis together with colitis. Moreover, liver-derived lymphocytes from patients with PSC show reduced expression of the IL-2 receptor and an impaired proliferative response to stimulation in vitro [62]. IL-2 can contribute to termination of inflammatory immune responses, by promoting the development, survival and function of T_reg; and a loss of IL2Rα signalling function in PSC is supported by the observation that patients who harbour variant polymorphisms exhibit reduced circulating T_reg populations [28]. A prominent role for TNFα in the immunopathogenesis of PSC has also been suggested through induction of immunopathogenic T-cell phenotypes as well as indirectly through the hepatic endothelial induction of mucosal chemokines and adhesion molecules that are normally ‘gut restricted’ in an NFκB-dependent manner [37]. Moreover, PSC genetic risk-associations include the 1p36 locus that encompasses the TNF-superfamily receptor TNFRSF14 – a protein expressed on CD4⁺ and CD8⁺ T-cells, B-cells, monocytes, neutrophils, dendritic cells and mucosal epithelium, which behaves as a molecular switch modulating lymphocyte activation [63].

An immunosuppressive role for histone deacetylase (HDAC)-7 – a gene implicated in the negative selection of T-cells in the thymus and development of tolerogenic immune responses – is supported by a genetic association at 12q13 in PSC GWAS, in which the most associated polymorphism was located within an intron encoding serine-threonine protein kinase (PRK)-D2 (19q13). When T-cell receptors of thymocytes are engaged, PRKD2 phosphorylates HDAC7 resulting in loss of its gene regulatory functions. This gives rise to apoptosis and negative selection of immature T-cells. Notably, this negative selection takes place owing to a loss of HDAC7-mediated repression of the leucocyte transcription factor Nur77 [56]. Nur77 expression parallels that of IL-10 and is heavily influenced by salt-inducible kinase (SIK)-2 polymorphisms, the latter of which is also proposed as a genetic risk locus in PSC.

Further impression of impaired mucosal tolerance is suggested through a genetic association at 18q21, which contains transcription factor (TCF)-4; congenital deficiency of which results in partial blockade of early B- and T-cell development and also attenuated development of immunoregulatory plasmacytoid dendritic cells (pDC) in murine models [64]. Caspase-recruitment domain (CARD)-9 is an important downstream mediator of signalling from mucosal PRR, and genetic associations suggest a link between defective intestinal mucosal microbial handling and the development of PSC. Card9 ^−/− mice appear more susceptible to experimentally induced colitis, typified by defective IFNγ and T_h17 responses, as well as reduced transcription of the mucosal chemokine CCL20, signifying the critical importance of CARD9 in the maintenance of epithelial immunostasis [65]. CCL20 has recently been identified as a GWAS risk locus in PSC [66] and as a chemokine facilitates the recruitment of T_h17 cells to epithelial structures. Moreover, CCL20 expression is under control of a positive feedback loop, wherein release of cytokines by T_h17 cells results in overexpression of CCL20 by BEC and intestinal epithelial cells in a paracrine manner [31, 41].

Another one of the strongest non-HLA associations in PSC is macrophage-stimulating (MST)-1, which is also associated with UC and Crohn’s disease (CD). MST-1 is expressed by BEC and involved in regulating innate immune responses to bacterial ligands, as well as modulating lymphocyte trafficking in lymphoid tissues through integrin- and selectin-mediated adhesion [67]. Glutathione peroxidase (GPX)-1 is an antioxidant enzyme located close to MST-1, and polymorphisms in GPX may also confer an increased disease susceptibility to PSC. Moreover, Gpx1/2 ^−/− mice develop a chronic ileocolitis with an increased frequency of colonic malignancy [68]; particularly noteworthy given the increased premalignant nature of colitis in PSC patients.

Variants in FUT2, an enzyme encoding galactoside 2-alpha-l-fucosyltransferase-2, have also been suggested to confer increased susceptibility to PSC (as well as Crohn’s disease), although fall short of reaching significance at a genome-wide level [69]. Fucosyltransferase variants alter the recognition and binding of various pathogens to carbohydrate receptors on the mucosal surface and are associated with changes in the commensal phyla in affected PSC patients characterised by elevated Firmicutes and reduced Proteobacteria. This is akin to changes found in FUT2 polymorphisms associated with Crohn’s colitis and again links defective immune responses to the gut microbiota in PSC. Moreover, variants in FUT2 have been described as a risk factor for development of dominant biliary stenosis in PSC – a phenotype associated with adverse clinical outcomes [70]. Colorectal malignancy may result as a direct consequence of altered fucosylation of the epithelial adhesion molecule E-cadherin [71], and a recent study in mice also illustrates how congenital E-cadherin deletion results in spontaneous periportal inflammation, periductal fibrosis and an enhanced susceptibility towards hepatobiliary cancer – akin to clinical PSC [72].

Collectively, genetic findings suggest that cholangitis in PSC may be a direct result of defective pathogen sensing, disrupted barrier functions, and askew in effector versus regulatory immune responses as part of a bigger picture where T-cell responses towards an hitherto unidentified specific antigen (or autoantigen) play a major role (Fig. 9.3).

Presently, there is no known medical treatment which is consistently demonstrated to attenuate disease progression in patients with PSC [79]. Consequently, clinical outcome is largely dictated by the development of cirrhosis, portal hypertension and variable predisposition to the development of colonic and/or hepatobiliary malignancy. Up to 50% of symptomatic patients develop progressive liver disease requiring transplantation, although modelling the natural history has proven challenging as a result of changes in diagnostic paradigms (magnetic resonance cholangiography now the ‘gold standard’), discrepant transplantation indications and phenotypic differences in inflammatory bowel disease globally. The absence of a defined serological marker makes case recognition difficult, and unbiased cholangiography assessments in patients with long-standing IBD found clinically manifest PSC in 2.7% of IBD patients compared with a genuine frequency of PSC-like lesions in up to 7.4% of the IBD population [80].

Despite the relatively low population frequency, chronic cholestatic liver diseases such as PSC remain a significant cause of morbidity and mortality for patients, accounting for a significant fraction of first liver transplantations in the Western world (up to 20–25% in Scandinavian areas) [81, 82]. As such, PSC portends a standardised mortality ratio (SMR) greater than fourfold that of a matched control population [83], although epidemiological registries illustrate significant discrepancy between event-free survival times across transplant centres versus true population-based cohorts (median 13.2 vs. 21.3 years [74]). Such study cohorts highlight the significant challenges in prognostic modelling, particularly in the unmasking of patients with early clinical disease, in addition to the inherent selection bias that stems from tertiary centre restricted reporting.

Distinct from the risk of progressive liver disease, patients with PSC harbour an increased predisposition towards hepatobiliary malignancy (Table 9.3), far beyond that posed by hepatocellular carcinoma (HCC) in the context of established liver cirrhosis. Cholangiocarcinoma (CCA) develops in up to 10–15% of patients. One-third of CCA are diagnosed within the first year of PSC diagnosis (annual incidence thereafter 0.5–1.5%); therefore, a heightened index of suspicion is advised for acute/subacute symptomatic PSC presentations, particularly men with persistently elevated bilirubin and coexisting colitis [84]. Abrupt changes in biochemical or clinical presentation of any PSC patient should also heighten suspicion of CCA. Biliary malignancy is cholangiographically indistinguishable from benign disease, although emerging endoscopic and molecular techniques may help discern malignant versus benign appearances (see Chap. 12: Gores et al.) [85]. Many centres in Europe and the USA apply biliary brush cytology with or without concomitant DNA analysis (by fluorescence in situ hybridisation [FISH], digital image analysis [DIA] or flow cytometry); however, the appropriate action upon pathological findings varies and lacks scientific justification [79]. Presently, however, no evidence-based screening protocols are available, and no current methodology facilitates effective identification of at-risk individuals prior to cancer development.

Gallbladder mass lesions are also more common in PSC with an estimated prevalence of 3–14%, with over 50% being established as adenocarcinomas at time of diagnosis [86]. Therefore, annual ultrasound surveillance is advocated in all patients to detect gallbladder polyps [87, 88], which if present warrant consideration for cholecystectomy.