The external triggers (bacteria translocation, bile toxicity, etc.) are likely to activate this aberrant adaptive immune system characterizing PSC, stimulating the transition into an activated phenotype of biliary epithelium
[26]. In chronic inflammation, cholangiocytes manifest a high proliferative capacity as an adaptative response to prevent ductopenia, and could also act as facultative liver stem cells in case of important liver damage
[27]. Inflammation also generates the expansion of the peribiliary gland compartment activating proliferation and hyperplasia of the biliary tree stem cells contained in the peribiliary niche
[28][29]. Cholangiocytes proliferation is extremally heterogeneous, involving two differentiated phenotypes
[30]: (1) the ductular reactive cholangiocytes, which seem to derive from the expansion of proliferative cholangiocytes and the peribiliary gland stem cells
[31][32] as described above (less likely, they could also derive from trans-differentiated hepatocytes)
[33]. This transition into reactive cholangiocytes has a crucial role in developing peribiliary fibrosis through the crosstalk with myofibroblasts, along the Hedgehog pathway
[28] and the hepatic stellar cells activation
[34], stimulating the excessive extracellular matrix deposition and fibrotic stricture development
[35]. (2) The senescent cholangiocytes, deriving from an extended DNA damage primarily involving N-Ras activation (with senescence-associated markers like p16(INK4a) and γH2A.x)
[36]. These cells share a senescence-associated secretory phenotype (SASP)
[37], with aberrant oncogenes and transcription factors (like ETS1) expression
[38] that promote apoptosis resistance, and secretion of several cytokines (i.e., IL-6, IL-8, CCL2, PAI-1) associated with the development of CCA
[36]. Previous studies have already described the presence of p53 dysfunction as the most common aberration in PSC-CCA, usually occurring in up to 30% of cases, more frequently than sporadic CCA
[39]. Also, presence of
ERBB2 amplification and
SMAD4 loss are more frequent in PSC-CCA
[40][41]. Other common mutations involve
KRAS,
GNAS, and
PIK3CA [40], as well as loss of chromosome 9p21 and inactivation of p16
[42]. Interestingly, some alterations like copy number variation, an indicator of chromosomal imbalance, occur earlier in PSC-CCA cancerogenic cascade
[40].
Among cytokines, IL-6 seems to have a mitogenic potential in CCA, promoting the expression on myeloid cell leukemia 1(McL-1) through STAT3 pathway, and concomitant inhibition of SOCS3
[43][44]. IL-6 also promotes PI3K/AKT pathway, resulting in antiapoptotic stimuli
[45]. An important role in cellular damage is played by biliary infections, promoting oxidative stress of cholangiocytes. Biliary infections activate the inducible nitric oxide synthase (iNOS) in inflamed tissue, with a consequent overproduction of nitric oxide (NO) and other reactive oxidants (e.g., 3-nitrotyrosine), which promotes direct DNA damage and inhibition of base excision repair
[46][47]. During infections, bacteria-derived lipopolysaccharide (LPS) also has a role in activating the epidermal growth factor receptor (EGFR) and the cyclooxygenase-2 (COX-2) and consequent stimulation of bile duct proliferation
[48], as well as the accumulation of some bile acids like deoxycholic acid, which could promote several mechanisms such as enhanced expression of interleukin IL-6 and downregulation of farnesoid X receptor (FXR)-dependent chemoprotection, which could facilitate cancer development
[49]. Genetic and epigenetic factors could contribute to PSC-CCA developments. For instance, natural killer cell receptor G2D (NKG2D) normally interacts with its ligand, major histocompatibility complex class I chain-related molecule A (MICA), with a protective role against tumor development. However, some polymorphisms of NKG2D seem to be associated with a lower receptor–ligand affinity and higher risk of PSC-CCA development
[50]. Among epigenetics changes, methylation of 9p21was seen to be a possible strategy of p16 silencing
[42]. Regarding microRNAs (miRNAs), their expression was seen to be associated with PSC and its progression to CCA. First observations came from mice models of chronic cholestasis. Induction of cholestasis was associated with acceleration of CCA development, which came with miRNAs down-regulation (miR-34a) or up-regulation (miR-210)
[51]. Voigtländer et al. found a significative expression of some miRNAs in the bile of PSC-CCA patients, most of them involved in the processes of epithelial–mesenchymal transition
[52]. Those miRNAs’ expression differs significantly in PSC and PSC-CCA, with higher levels in the second group. Interestingly, the levels of this specific miRNA were substantially lower in patients with CCA alone, suggesting a different etiopathogenesis in CCA developing on an underlying biliary disease.
3. Surveillance of CCA in PSC
The silent development of this neoplasia, typically asymptomatic in early stages, is in contrast with its aggressive behavior and rapid progression. This characteristic makes it difficult to achieve early diagnosis, losing the therapeutical window for curative treatment and compromising patient outcomes, due to limited systemic treatment options. In PSC patients, the radiological surveillance has a crucial role to detect early CCA and to allow curative treatments, such as liver resection or transplantation in patients who are usually young and fit for surgery. An ideal surveillance strategy should be simple, cheap, accepted by the patient, non-invasive, and with high sensitivity, aiming to increase overall survival. However, no diagnostic tests have those characteristics
[53], and high-grade dysplasia or early-stage CCA are usually diagnosed using invasive techniques like ERCP, keeping in mind that the risk of complications in ERCP is not justified by the sensitivity of the endoluminal sampling, that is usually low
[54]. The positive impact of radiological surveillance for PSC patients has already been proven by different studies, reporting better outcomes with a reduction in hepatobiliary cancers-related deaths in patients exposed to at least a routinary imaging technique
[55][56]. Also, the large multi-center study from the IPSCSG proved that scheduled imaging was associated with improved overall survival
[53]. Magnetic resonance imaging (MRI) with magnetic resonance cholangiopancreatography (MRCP) is the preferred radiologic modality with no radiation and the possibility to supply more detailed and accurate information than abdominal (US)
[57]. The guidelines suggested yearly MRI/MRCP for CCA surveillance in large-duct PSC patients. Pediatric and sdPSC patients represent low-risk categories, and surveillance is not strictly indicated for them
[58][59].
There is not a consensus on the correct timing of scheduled imaging for CCA surveillance in PSC, and some experts suggested a limited role of imaging surveillance for biliary malignancies
[60]. Recently, a prospective study from Villard et al., including 512 unselected patients followed for 5 years with annual MRI/MRCP, detected CCA in 2% of patients, associated with previous development of progressive and severe biliary strictures. Although 62% of them were eligible for curative treatment, all of them except one developed tumor recurrence and died during the follow-up, pointing out the inability of annual surveillance to detect cancer early enough to provide long-term survival
[61]. Moreover, the failing of the guidelines’ strategy has been evident in the group of patients with severe strictures, suggesting how the future direction of research should be the individualization of surveillance, reserving closer MRI/MRCP controls in selected high-risk groups (recent or long-date diagnosis of large-duct PSC, older age, or severe and symptomatic disease).
The use of scheduled ERCP associated with MRI/MRCP does not improve the surveillance efficacy
[53][62], and PSC itself represents a risk factor for ERCP-associated complications
[63]. Scheduled ERCP in PSC could have a beneficial effect in the treatment of high-grade stenosis, regardless of patient symptoms according to recent data
[64], but does not represent an adequate tool for surveillance.
Abdominal US remains the cornerstone for the surveillance of the other types of hepatobiliary neoplasms occurring in PSC; surveillance of GBC requires a 12-month US surveillance, which could be replaced with or added to the annual MRI/MRCP for CCA surveillance if already performed. A 6-month US is mandatory in the presence of cirrhotic evolution for HCC surveillance
[58][59].
The majority of PSC international guidelines discouraged the 6-month use of CA19-9 for CCA surveillance
[58][59][65][66] because it has a low positive predictive value and a low accuracy
[67]. Indeed, CA19-9 is highly variable and many other benign or malignant conditions, as well as interpersonal variability, could influence the serum levels of these markers. However, the rapid increase of CA 19-9 has been shown to predict malignancies regardless of the cut-off in use, and some experts have tried to narrow the interval of assessment of the biomarker (every 2 or 3 months) in order to detect the progressive CA19-9 change rather than focus on its absolute value
[68]. The risk of over-diagnosis associated with this strategy has not been evaluated yet, and the potentially higher rate of false positives could remarkably affect the costs of the surveillance program.
4. Future Perspective: Next-Generation Biomarkers for PSC-CCA Diagnosis and Surveillance
The pancreatic autoantibody against glycoprotein 2 (anti-GP2 IgA) had been indicated as a novel marker in large-duct PSC. It represents a prognostic marker associated with a more severe disease and dismal prognosis with a higher risk of developing CCA
[69]. Another observation on biochemical markers reveals that a high enhanced liver fibrosis (ELF) score correlates with patients developing CCA sporadically or on PSC. A following multivariate analysis also proved that ELF score was correlated to CCA independently of liver disease stage or patients’ age, suggesting this marker to be related to the biological nature of tumor development
[70]. This follows that some markers could be a useful tool for patient risk stratification, being able to identify patients with a higher risk of developing PSC-CCA. This could guide to a more individualized surveillance strategy in PSC. A recent study from Cuenco et al. analyzed the diagnostic performance of a panel of new biomarkers including pyruvate kinase M2 (PKM2), cytokeratin 19 fragment (CYFRA21.1) and mucin 5AC (MUC5AC) achieving promising results with an high specificity (90%) but mostly a high sensitivity when combined with gamma-glutamyltransferase (GGT) (82%)
[71]. Also, osteopontin, usually upregulated in CCA tumor cells, was seen to be higher in the serum of patients with CCA compared to healthy controls. During the exploratory analysis of the study from Loosen et al., some 10 PSC patients and 13 healthy controls were compared with 27 patients with CCA, revealing significantly elevated osteopontin levels in CCA patients compared to serum samples from healthy and PSC patients, suggesting a possible role in differentiating among PSC patients those who develop biliary malignancy
[72]. Unfortunately, no PSC patients were included in the final analysis, and the observation remains limited to the small exploratory cohort.
A study on the use of volatile organic compounds (VOCs) measurement in bile showed interesting results, revealing differences in gas concentrations between PSC and PSC-CCA patients
[73], as already observed in other types of cancer
[74]. The subsequent study from the same group of Navaneethan et al. also showed accuracy of VOCs in distinguishing CCA from PSC patients when measured in urine. Some VOCs like ethane and 1-octene reached an 80% sensitivity at 100% specificity, and also 2-propanol and acetonitrile showed promising results, representing a potential non-invasive marker. In the future, the possibility of translating VOCs measurement to breath analysis might be tested
[75].
DNA hypermethylation of gene promoters represents an early event in biliary carcinogenesis. The methylation of CDKN2A (p16) represents a frequent DNA alteration
[76], but also other loci were found to be hypermethylated in CCA samples
[77][78][79], representing potential markers for PSC-CCA detection. However, only a few studies included PSC patients, and data on PSC-CCA are limited. The analysis of methylation status of some candidate genes (cysteine dioxygenase type 1 (CDO1), cannabinoid receptor interacting protein 1 (CNRIP1), septin 9 (SEPT9), and vimentin (VIM)) on biliary brushing of CCA and PSC patients revealed good performances in discriminating malignancy from the benign and inflammatory-related biliary changes in PSC, with high sensitivity (85%) and specificity (98%)
[80]. A more recent study tried to validate the measurement of methylation level of the same gene panel in bile, using droplet digital PCR (ddPCR) for the analysis
[81]. Previous studies on bile samples could not find a significative association between hypermethylation of specifical genes (CDKN2A) and PSC-CCA
[82], or measured potential biomarkers only in non-PSC patients (like CCND2, CDH13, GRIN2B, RUNX3, and TWIST)
[83]. Instead, the three center-based (Sweden, Norway, Finland) study from Vedelet et al. observed the gene methylation panel including CDO1, CNRIP1, SEPT9, and VIM to be accurate for early PSC-CCA detection, with an AUC of 0.88. The results were even enhanced including only patients with early PSC-CCA diagnosis (≤12 months) and PSC controls with long follow-up (>36 months), reaching a sensitivity and specificity of 100% and 93%, respectively, and an AUC of 0.98. This promising result proved that DNA methylation biomarkers in bile has the potential to complement standard diagnostic modalities for early PSC-CCA detection
[81]. Prospective validation of this study in larger multicenter sample series is warranted. However, the possibility to include this methylation analysis in CCA surveillance in PSC patients remains controversial due to the invasive nature of bile sampling. A few studies also performed DNA methylation analysis on blood samples, discovering some potential biomarkers, already tested in tissue samples
[84] (SHOX2, SEPT9, OPCML, HOXD9, etc.), but unfortunately no PSC patients had been included
[85][86]. The analysis of micro ribonucleic acid (miRNAs) was highly investigated. miRNAs represent non-coding small nucleic acids capable of binding to specifical messenger RNA (mRNAs) and regulating gene expression
[87][88]. Several miRNAs share an important role in chronic cholestasis, interfering with several cellular processes including proliferation and apoptosis. Impaired levels of miRNAs could result in dysregulating cellular cycle generating clonal expansion and eventually cancer. Interestingly, miRNAs expression is not directly altered by cholestasis; indeed, presence of high levels of bilirubin does not influence the result of miRNAs analysis
[52]. As already explained above, bile levels of some miRNAs (miR-640, miR-3189, miR-1537, and miR-412) are significantly higher in PSC-CCA patients compared with PSC-alone patients, representing a potential diagnostic tool in early diagnosis. Another study from Li et al. aimed to investigate potential biomarkers for CCA diagnosis regardless of PSC, showing the diagnostic value of an miRNA panel specifically associated with CCA with good sensitivity of 67% and high specificity of 96%. The study included a few PSC patients in the control group, suggesting their possible role in differentiating the miRNA pattern
[89]. This aspect needs to be considered in future studies. Some differences were found between miRNAs levels of PSC and CCA-alone patients (miR-1281, miR-126, miR-26a, miR-30b, and miR-122). Nevertheless, a consistent overlapping expression of most miRNAs in PSC and CCA patients was observed. Only miR-126 gave considering results, but only reached a specificity of 93% with poor sensitivity, comparable with brushing cytology
[52]. However, another study of miRNA in serum showed a difference in the expression of miR-222 and miR-483-5p in CCA versus PSC controls
[90]. Other miRNAs (miR-21, miR-221, miR-122, miR-192, miR-29b, and miR-155) were demonstrated to be associated with CCA in terms of both diagnosis and prognosis; however, control groups of these studies count only a few or no PSC patients
[91][92]. Future research is required to investigate on serum miRNAs detection, being less invasive compared to bile sampling. Combined strategies should be implemented to include more information from both serum and bile and integrate them with other biomarkers.
Neoplastic cells are characterized by high metabolism with production of several potential biomarkers. Indeed, metabolic alteration could reflect on impaired concentration of circulating metabolites in the presence of cancer. Specific alteration at the metabolome analysis had been already described in other neoplastic diseases
[93][94]. According to that, the group of Banales et al. performed a metabolomic analysis of 424 different metabolites in serum to identify a significant concentration of certain circulating metabolites in patients who developed hepatic malignancy, in particular iCCA or hepatocellular carcinoma (HCC) compared with PSC. Some metabolites were significantly different in the PSC and iCCA group and the algorithms combining some of them, histidine and PC(34:3), could accurately differentiate between the two diagnoses
[95]. This suggests a different metabolic profile of CCA and PSC that could represent a starting point for further analysis on PSC-CCA patients.
Promising results also came from proteomic analysis. In particular, urine proteomic analysis reveals the capacity to differentiate CCA from PSC
[96]. The use of urine proteomic analysis showed better results when associated with biliary proteomic analysis, which also provides good discrimination capability in the field
[97]. The combination of bile and urine proteomic analysis had been explored in a retrospective study and then applied to a prospective analysis, improving sensitivity of CCA diagnosis, compared with standard diagnostic approach, from 72% to 94%. A large prospective trial on patients with progressive cholestasis during surveillance is required to confirm the accuracy of this method for the early diagnosis of CCA
[98].
In serum, a glycomic and proteomic analysis on a particular set of proteins was able to identify some differences in the expression of certain glycans in patients with CCA. Among them, fucosylated fetuin A (fc-fetuin A) was capable of differentiating CCA from those with PSC, suggesting a possible role of this analysis in the surveillance of PSC patients
[99].
An early field of liquid biopsy-based research is the analysis of extracellular vesicles (EVs) and their proteomic profile in serum samples. EVs contain biomolecules of various nature, like lipids, proteins, and nucleic acids, and different cells, including malignant cholangiocytes, release EVs with specific possible biomarkers
[100][101]. First evidence of specifical EVs proteomic patterns came from the analysis of tissue samples, defining several CCA-derived EVs, not in common with PSC, that may reflect tumor features and give prognostic information
[102]. Lapitz et al. performed an analysis of EVs’ transcriptomic profile in urine and serum of CCA, showing a specific expression of messenger RNAs (mRNA) and non-coding RNAs (ncRNAs) in CCA patients compared to a mixed group of patients with PSC, UC, and healthy controls. In urine, in the partial analysis including only CCA (
n = 23) vs. PSC (
n = 5) patients, a higher expression of some specific mRNAs like CLIP3, VCAM1, and TRIM33 in CCA group (AUC 0.965) was reported, as well as a significant different expression of other transcripts like ATP5EP2, LOC100134713, and SNORA8
[103]. The same analysis in serum (CCA
n = 12 vs. PSC
n = 6) reveals mRNAs, PON1, ATF4, and PHGDH were the best candidate biomarkers for the differential diagnosis of CCA and PSC (AUCs 1.00), as well as a high accuracy for the identification of CCA vs. PSC of the transcripts MALAT1, LOC100190986, and SNORA11B (AUCs: 1.00)
[103]. Subsequent gene ontology analysis indicated that most of the commonly altered mRNAs participate in carcinogenic pathways. Among EVs biomarkers, proteins also have a differential expression in CCA or PSC (FIBG, A1AG1, S10A8) according recent studies, and interestingly some of them have higher diagnostic values in early-stage CCA (I-II) (FCN2, ITIH4 and FIBG)
[102]. A recent proteomic analysis of EVs in serum from the same study group individuates the combination of serum biomarkers CRP/FIBRINOGEN/FRIL as a valid tool for diagnosis of early CCA developing in PSC patients. Other biomarkers were associated with a prognostic role, predicting overall survival. Moreover, the integration of serum PIGR with CRP/FIBRINOGEN/FRIL showed the capacity for predicting CCA development in PSC patients with sensitivity and specificity of 64% and 93%, respectively (AUC 0.91), suggesting its potential use in surveillance settings in the future
[104]. Anyway, further studies are needed to validate their effective role and the potential therapeutic strategies associated with their measurement.
Next-generation sequencing (NGS) provides a multigene analysis with high analytical sensitivity and recently grew up with the advances in the characterization of the mutational profile of biliary tumors. Thanks to NGS, many recurrent mutations had been already identified in CCA
[105], and some studies already tried to apply NGS for the research of novel biomarkers in de novo CCA diagnosis. A prospective study from Singhi et al. evaluated the use of NGS (a 28-gene panel renamed BiliSeq) in bile duct brushing and tissue sample obtained during ERCP. There was an improvement of sensitivity compared to standard clinical and pathological evaluation (73% vs. 48%). Notably, sensitivity of the BiliSeq panel was considerably higher (83%) in patients with underlying diagnosis of PSC
[106]. More recently Arechederra et al. evaluated the use of NGS for the mutational analysis of bile cell-free DNA (cfDNA). The panel in use, called the “Bilemut assay”, achieved a sensitivity of 96.4% and a specificity of 69.2%
[107]. The use of bile cfDNA could provide the advantage of obtaining nucleic acids released from all the CCA cells along the biliary tract, overcoming the limitation of the tissue sample and improving sensitivity. In general, the use of NGS for the identification of biomarkers could be useful not only for the early detection of malignancy but also for its prognostic role, being able to detect mutations that could become potential targets of oncological therapies. Further research will fill the gap made by the absence of specifical studies on PSC-CCA patients, testing the diagnostic and prognostic performances of NGS biomarkers panels in this group of high-risk patients.