Biomarkers of intrahepatic cholangiocarcinoma: diagnosis and response to therapy

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver cancer behind hepatocellular carcinoma (HCC) and carries a dismal prognosis. Improved genetic analysis has paved the way for a better understanding of the distinct somatic genomic landscapes of ICC. The use of next generation sequencing has paved the way for more personalized medicine through identifying unique mutations which may prove to be therapeutic targets. The ability to identify biomarkers specific to ICC will assist in establishing a diagnosis, monitoring response to therapy, as well as assist in identifying novel therapies and personalized medicine. Herein, we discuss potential biomarkers for ICC and how these markers can assist in diagnosis, monitor response to therapy, and potentially identify novel interventions for the treatment of ICC.

ICC is thought to occur due to chronic inflammation that can lead to an inflammatory milieu that damages DNA and induces cholangiocyte proliferation [9,10]. Another theory of ICC pathogenesis hypothesizes that hepatic progenitor cells overexpress Notch1 with oncologic transformation through a cholangiocellular pathway [11,12]. Risk factors for ICC are well established and include cirrhosis, viral hepatitis, primary sclerosing cholangitis (PSC), parasitic infections, carcinogen exposure, as well as several genetic syndromes such as Lynch syndrome, BRCAassociated protein-1 (BAP-1) tumor predisposition syndrome, cystic fibrosis, and biliary papillomatosis [13]. Additionally, ICC exists in two predominant subtypes: proliferative and inflammatory [14]. The proliferative subtype, characterized by activation of oncogenic signaling pathways, DNA amplifications, and mutations in BRAF and KRAS, tends to be more poorly differentiated than the inflammatory subtype, characterized by activation of inflammatory signaling pathways and overexpression of cytokines, and is associated with a worse prognosis [14].
Surgical resection represents the only potentially curative treatment for patients with ICC and unlike HCC or PHCC, liver transplantation is not an established treatment option [15][16][17][18][19][20]. However, approximately only one third of patients present with resectable disease and, despite surgical resection with negative margins, early disease recurrence is common [21]. Even when patients undergo surgical resection with curative intent, 5-year overall survival (OS) is only 20-35% [22]. While chemotherapy may prolong survival for select patients, drug resistance and significant toxicities, especially in patients with poor performance status, limit the success of systemic therapy [23].
Through improved genetic analysis there is a better understanding of the distinct somatic genomic landscapes of biliary tract cancers such as ICC [24]. The use of next generation sequencing has paved the way for more personalized medicine through identifying unique mutations which may prove to be therapeutic targets. The ability to identify biomarkers specific to ICC will assist in establishing a diagnosis, monitoring response to therapy, as well as assist in identifying novel therapies and personalized medicine [25]. In this review, we discuss potential biomarkers for ICC and how these markers can assist in diagnosis, monitor response to therapy, and potentially identify novel interventions for the treatment of ICC.

Diagnostic biomarkers of intrahepatic cholangiocarcinoma
When evaluating a liver lesion, it is important to distinguish ICC from other liver tumors such as HCC or metastatic disease. The diagnosis of ICC is primarily made with imaging in the appropriate clinical context and, if needed, subsequent biopsy [26]. Computed tomography (CT) scan and/or magnetic resonance imaging (MRI) are the imaging modalities of choice for ICC. Contrast enhancement patterns on imaging may be able to distinguish ICC from HCC [27]. ICC receives its blood supply from the portal vein and thus often has a portal or delayed phase enhancement pattern while HCC receives its blood supply from the hepatic arteries and displays an arterial phase enhancement pattern on CT [28,29].
Several biomarkers have been identified to assist in the diagnostic work up of liver lesions (Fig. 1). Tumor markers provide useful diagnostic and prognostic information as adjunct confirmatory tests during the workup of solid tumors and in the postoperative surveillance setting to help monitor for recurrent disease; however, tumor markers should not be used as the sole means of diagnosis [30].
The diagnosis of intrahepatic cholangiocarcinoma (ICC) is primarily made with imaging and subsequent biopsy as it is important to distinguish ICC from other liver tumors such as hepatocellular carcinoma (HCC). Biomarkers have been identified in the blood, bile, and tumor tissue to assist in the diagnostic work up of liver lesions.
Carbohydrate antigen (CA) 19-9, a well-known biomarker for CCA, is a Lewis blood group antigen produced by pancreatic, biliary ductal, gastric, and colonic epithelial cells [25]. However, CA19-9 is not produced, and therefore not detectable, in 7% of the population [31,32]. An elevated CA19-9 is more commonly associated with ICC, as opposed to an elevated α-fetoprotein (AFP) that is more suggestive of HCC [33]. CA19-9 has a 72% sensitivity and 84% specificity as a diagnostic biomarker for CCA. Indeed, CA19-9 can be elevated in benign conditions such as biliary obstruction, cholangitis, and primary biliary cirrhosis.
PSC is an idiopathic, cholestatic liver disease characterized by persistent progressive biliary inflammation and fibrosis and is a risk factor for bile duct cancers [34]. In one study, a CA19-9 value of 129 U/mL demonstrated a 79% sensitivity and 98% specificity for CCA in patients with PSC [35]. In a different study, Vedeld et al. [36] investigated the utility of DNA methylation biomarkers in bile for early diagnosis of CCA in patients with PSC. Using droplet digital PCR (ddPCR), the authors analyzed 344 bile samples from 273 patients with sporadic and PSCassociated CCA, as well as other non-malignant liver disease for promoter methylation of CDO1, CNRIP1, SEPT9, and VIM. All four markers were associated with CCA detection among patients with PSC up to 12 months before conventional CCA diagnosis.
In addition, activating mutations in the oncogenes EGFR (ErbB1), HER2 (ErbB2), and PDGFa as well as silencing the tumor suppressor genes TP53 and CDKN2A may play a role in the pathogenesis of CCA [37]. These aberrations often correspond with the gain of chromosomal fragments 5q, 7p, 8q, 17q, and 20q and loss of 3p, 6q, 9p, and 17p [37,38]. Interestingly, patients who develop CCA in the setting of a liver fluke infection gain chromosomal fragment 21q22 and lose fragments 1p36, 9p21, 17q13, and 22q12 [39]. KRAS has been reported to be one of the most frequently mutated genes in ICC and may serve as a potential biomarker [40][41][42]. KRAS, as well as BRAF mutations, are present in approximately 10% of patients with ICC and approximately 30% of patients harbor mutations in the PI3K/PTEN/AKT/mTOR signaling pathway [43,44].
Biomarkers may also help differentiate malignant disease from benign biliary disease. FAM19A5 and RBassociated KRAB zinc-finger protein (RBAK) were elevated in patients with CCA versus patients with benign biliary conditions [53]. Serum α1b-Glycoprotein (A1BG)/afamin (AFM) ratio greater than 1.8 differentiates patients with CCA from healthy patients with a 84.4% sensitivity and 87.5% specificity [54]. Serum and bile levels of Wisteria floribunda agglutinin (WFA) may also differentiate ICC from benign biliary diseases [55].
Bile based markers have also been used as a diagnostic aid. The fluid sample is closer to the presumed tumor tissue, a potential benefit, but sample collection requires an invasive procedure [25]. Patients with CCA have significantly lower total bile concentrations and deoxycholic acid ratios than patients with benign biliary disease due to altered bile acid transport [56]. Sperm specific protein 411 (SSP411) is elevated in the bile of CCA patients compared with patients who have benign biliary disease [57].

Biomarker predictors of outcomes
Outcomes for patients with ICC are generally poor, and recurrence is common after resection. Additionally, a hepatectomy is a physiologically demanding procedure and patient selection is important to achieve operative success [26]. As such, biomarkers to assist in risk stratification and guide treatment decisions is an active area of investigation ( Table 1). As previously discussed, CA19-9 is a useful biomarker in the diagnosis of ICC. In addition, CA19-9 may have prognostic significance for patients with ICC. Moro et al. [65] demonstrated that preoperative CA19-9 and CEA were prognostic of OS when a cutoff of 176.3 IU/mL for CA19-9 and 9.6 ng/mL for CEA were utilized. Other studies indicate that CA19-9 is elevated in 57% of patients with ICC, and a CA19-9 level higher than 37 U/mL was predictive of lymph node metastasis and survival [66,67]. Among patients with ICC who underwent hepatectomy, Qiu et al. [68] reported that a low aspartate aminotransferase (AST) to lymphocyte ratio index combined with a low CA19-9 level was associated with better OS and disease-free survival (DFS). Additionally, 237 patients with ICC who had undergone resection had arginase-1 and glypican-3 assessed via immunohistochemistry, and high arginase-1 and glypican-3 expression was associated with a poor prognosis [69].
Tsilimigras et al. [70] developed the LabScore scoring system based on data from 660 patients who underwent hepatectomy for ICC. The LabScore includes platelet count, CA19-9, albumin, and the neutrophil-to-lymphocyte ratio (NLR). A higher LabScore was associated with worse tumor characteristics, TNM stage of disease, and was significantly associated with 5-year OS, DFS, and disease recurrence. Furthermore, Tsilimigras et al. [71] created a classification tree based on the analysis of 826 patients with a history of ICC resection and divided them into 3 clusters: common, proliferative, and inflammatory according to tumor size, CA 19-9, and NLR. Although patients in the inflammatory cluster had the lowest CA19-9 levels, midsized tumors, and the highest NLR, these individuals had the worst median OS.
KRAS and BRAF mutations may also be associated with prognosis [43,44]. KRAS mutations are associated with perineural invasion and a worse post-operative survival in patients with ICC [72]. However, patients with KRAS mutations had a worse 5-year OS than patients with BRAF mutations (13.5 vs 23.2 months) [43]. Additionally, elevated EGFR was associated with a worse prognosis with a shorter median OS (8.5 months versus 38.5 months) [73], and reduced PTEN expression is a predictor of poor OS in patients with ICC who have undergone resection [40,74]. TP53 mutations have a prevalence of 0.7-37% in patients with ICC and are generally associated with a worse prognosis [75,76]. MET mutations occur in approximately 12-58% of ICC tumors, and MET overexpression is associated with increased invasion and poor prognosis [77,78].
DNA methylation, histone modification, and noncoding RNA-associated gene silencing may initiate and sustain epigenetic changes involved in the pathogenesis of ICC [25]. Additionally, micro-RNA expression patterns are involved in the pathogenesis of ICC and can differentiate tumor from normal tissue [79]. Alternative splicing is a critical step in post-translational modification of mRNA and can predict the prognosis and recurrence of HCC and ICC [80][81][82][83]. A cluster analysis based on differentially expressed alternative splicing (DEAS) was performed with HCC, ICC, and normal liver tissue [84]. Luo et al. [84] reported differences in DEAS between the samples and highlighted the prognostic significance of DEAS among the tissue samples, developing predictive models that demonstrated clinical utilization. Additionally, increased levels of the heat shock protein 70-kDa protein 1 (HSP70.1), involved in regulating the cell cycle, may be inversely correlated to OS in patients with CCA [85,86]. Tumor type M2 pyruvate kinase (TuM2-PK) can be useful to distinguish CCA from benign disease as levels are often elevated in CCA proportional to tumor burden such that high levels of TuM2-PK are seen in patients with lymph node metastasis [87][88][89]. In one study, the sensitivity and specificity of TuM2-PK for CCA exceeded that of CA19-9 and was able to discriminate CCA from healthy controls [25,88]. Connective tissue growth factor (CTGF) expression may be associated with longer DFS and OS, but the mechanism of how CTGF influences tumor biology remains largely unknown [90]. Wnt1-inducible signaling pathway protein 1 (WIPS1), part of the WNT pathways, is involved in regulating cell proliferation, differentiation, adhesion, migration, and survival. WISP1 expression is associated with ICC carcinogenesis, overexpressed in 49% of ICC cases, and associated with a poor prognosis [91,92]. Elevated CYFRA 21-1 was associated with poor 3-year RFS and OS [93]. MMP-7 is expressed by malignant cholangiocytes and predicts poor post-operative survival. Similarly, MMP-9 predicts lymph node metastasis [94].
Core 3 synthase plays an important role in the digestive system, and cells expressing core 3 synthase show lower migratory and invasive rates, as well as lower metastatic activity. Indeed, in CCA, the expression of core 3 synthase, identified by the antibody G8-144, was associated with lower mortality rates [105]. On the other hand, expression of 6-sulfated N-acetyllactosamine on the extended core-1 O-glycans, identified by the antibody MECA-79, was associated with an unfavorable prognosis [105]. Additionally, dual-specificity phosphatase 11 (DUSP11) was evaluated in eight pairs of ICC, PHCC, and distal CCA, and their corresponding adjacent tissue by qPCR. In all types of CCA, DUSP11 was elevated compared with the adjacent tissue. In ICC, high DUSP11 was associated with an advanced T stage and poor prognosis, which was not the case for ECC or PHCC [106]. DDK1 expression in tumor tissue from ICC is associated with elevated MMP-9 and vascular endothelial growth factor-C (VEGF-C) expression which, in turn, is associated with tumor invasion and a high incidence of lymph node metastasis [107]. Additionally, IDH mutations were more common in tumors with poor histology and are associated with worse survival after resection [50].

Biomarkers to guide therapy
Systemic chemotherapy options for ICC are limited as drug resistance and drug-related toxicities are common [23]. Gemicatine with cisplatin is the standard systemic therapy for advanced cholangiocarcinoma [108]. However, through better genetic analysis and understanding of distinct biomarkers, targeted therapies have been developed in an attempt to improve treatment response and survival [24,25]. As previously discussed, IDH mutations may be present in 15-30% of patients and are a poor diagnostic marker [50,51]. IDH inhibitors have been utilized in the treatment of ICC with limited success. The ClarIDHy study was a phase III randomized controlled trial (RCT) involving 185 patients with IDH1-mutated cholangiocarcinoma. The patients were assigned to the IDH1 inhibitor, ivosidenib, or placebo. Patients who received ivosidenib had significantly longer PFS compared to placebo (2.7 versus 1.4 months, respectively) [109].
Lapatinib, an inhibitor of EGFR and HER-2, as well as trastuzumab, a HER2 inhibitor, have some demonstrated efficacy in CCA [110]. Similarly, among patients with advanced biliary cancer, including ICC, erlotinib, an EGFR inhibitor, with or without bevacizumab, a VEGF inhibitor, has clinical efficacy in ICC [111,112]. However, the addition of erlotinib to GEMOX did not improve OS or PFS [113].
Fibroblast growth factor receptor (FGFR) alterations are present in 10-15% of patients with ICC [52]. Pemigatinib, an inhibitor of FGFR, has been reported to have a 35% objective response rate in ICC patients harboring an FGFR mutations [114]. However, 42% of patients died from disease progression and 45% of patients had serious adverse events [114]. Nevertheless, pemigatinib is currently under investigation in a phase 3 RCT that compares the efficacy and safety of pemigatinib versus gemcitabine and cisplatin among patients with advanced or metastatic cholangiocarcinoma with an FGFR2 fusion or rearrangement (NCT03656536) [115]. Similarly, futibatinib, a different inhibitor of FGFR, was investigated in patients with ICC and FGFR2 alterations who had disease progression after first-line therapy [116]. Futibatinib had a 34% objective response rate and 76% disease control rate at ≥6 months follow-up, but serious adverse events occurred in 73% of patients [117]. Futibatinib is currently being compared with gemcitabine-cisplatin in a phase 3 study of patients with advanced cholangiocarcinoma and an FGFR2 fusion or rearrangement (NCT04093362) [115].
A study by Chen et al. [118] performed targeted next generation sequencing in 98 patients with advanced biliary tract cancer treated with camrelizumab plus gemcitabine and oxaliplatin. The authors noted that KRAS and TP53 mutations were much more frequent in advanced-stage biliary tract cancers than in early-stage disease. KRAS-TP53 co-mutations were favored in advanced CCA, with a favorable response to immunotherapy and single KRAS mutations predicted poor prognosis and immunotherapy outcomes for CCA. Tsilimigras et al. [119] reported on the role of tumor burden as a predictor of outcomes in 1101 patients with ICC who received surgical treatment. Patients were divided into groups of low, medium, and high tumor burden. The 5-year OS was incrementally worse as the tumor burden increased. In subgroup analysis, patients with high tumor burden that received adjuvant chemotherapy had significantly better outcomes than individuals who did not.

Immune signature and ICC
Immune based therapies have changed the landscape of cancer care from directly targeting the tumor itself to manipulation and activation of the immune system to eradicate tumor cells. However, a minority of patients respond to immunotherapies and advances are needed in immune based biomarkers to predict response to therapy and guide treatment decisions. Immune checkpoint inhibitors (ICIs) are currently approved for patients with solid gastrointestinal malignancies that have mismatch repair deficiency that includes ICC [120,121]. Unfortunately, mismatch repair deficiencies are reported in only 1-10% of patients with ICC [122]. Mismatch repair deficiencies were detected more frequently among patients with liver-fluke associated tumors [58]. Pembrolizumab, an anti-PD-1 ICI, is currently approved for the treatment of solid tumors with mismatch repair deficiencies [120]. Nivolumab, another anti-PD1 ICI, and pembrolizumab have demonstrated acceptable response rates in early phase clinical trials in patients with biliary tract cancers [123][124][125][126]. The LEAP-005 study is currently investigating lenvatinib, a kinase inhibitor, in combination with pembrolizumab in patients with advanced solid tumors. Interim analysis has demonstrated an overall response rate of only 10% with a disease response rate of 21% and duration of response of 5.3 months [127].
PD-L1 expression may be present in up to 70% of ICC tumors and be associated with worse survival [122,128]. PD-L1 expression correlates with response to ICIs in patients with non-small cell lung cancer, gastric cancer, and urothelial cancer, but efficacy data is limited among patients with ICC [122,[129][130][131]. There has been increased interest in the possible utility of DNA damage repair (DDR) gene mutations as a predictive biomarker to immunotherapy response. DDR gene mutations, such as in poly (ADPribose) polymerase 1 and 2 (PARP) or breast related cancer antigens (BRCA), prevent the ability of cells to repair DNA damage effectively repair DNA damage resulting in genomic instability [122,132]. DDR deficiency may lead to antitumor immunity by activating the innate immune response [133]. Additionally, other immune or inflammatory markers such as C-reactive protein (CRP) are associated with tumor recurrence after resection. For example, a CRP level <1.0 mg/dL was a favorable prognostic factor among patients with biliary tract cancers receiving chemotherapy [134]. Furthermore, interleukin-6 (IL-6) is proportional to pre-operative and post-operative tumor burden in patients with CCA [135]. Therefore, IL-6 may be useful as a potential diagnostic marker but low specificity limits its utility in this manner [136,137]. Transforming growth factor (TGF)β plays a role in cancer development as it is essential for cellular proliferation and differentiation. The expression of TGF-β is an indicator of early tumor recurrence [138]. Similarly, SMAD4, a protein involved in TGF-β signaling, is downregulated in approximately 55% of patients with ICC and associated with increased lymph node metastasis and poor tumor differentiation [139,140].
Recent data have noted four immune subsets of ICC characterized based on the composition of the tumor microenvironment. The immune desert phenotype is the most common, comprising 48% of cases, and is characterized by weak expression of immune markers. Meanwhile, the immunogenic pattern is characterized by a high amount of innate and adaptive immune cells and strong activation of inflammatory and immune checkpoint pathways. The myeloid rich subset is characterized by moderate to strong expression of myeloid signatures and a low lymphocytic signature. The last subtype has mesenchymal features with strong expression of activated fibroblast and is most associated with a poor prognosis [141].
The use of adoptive cell transfer may benefit patients with metastatic cholangiocarcinoma. Tran et al. [142] utilized whole-exome-sequencing to demonstrate that tumor infiltrating lymphocytes (TIL) from a patient with metastatic cholangiocarcinoma contained CD4+ T cells that recognized a mutation in erbb2 interacting protein (ERBB2IP) and the transfer of TIL decreased metastatic tumor burden. In addition, upon disease progression, the patient was again treated with TIL and experienced disease regression. This case report demonstrated the utility of sequencing to target unique mutations to provide a highly personalized therapy for patients with advanced disease. Evolving immune therapies may show promising results in the treatment of ICC (Table 2).

Future directions
Biomarkers assist in the diagnosis of multiple malignancies and emerging data support their use to guide patient selection for both surgery and systemic therapy. An important pathway to improve survival for patients with ICC is to identify biomarkers that appropriately select patients who might benefit from either neoadjuvant or adjuvant systemic therapy, as well as identify patients with tumors at high risk    for recurrence after resection. While systemic therapy provides a modest benefit to some patients with ICC, other patients may be unlikely to enjoy therapeutic benefit while experiencing toxicity. As such, the appropriate selection of patients may help avoid morbidity and minimize unnecessary exposure to toxic therapies for patients unlikely to derive clinical benefit. Currently, gemcitabine in combination with cisplatin remains the standard systemic treatment for advanced ICC with a median OS of 11.7 months versus 8.1 months among patients receiving gemcitabine alone [108]. Additional studies of systemic and locoregional therapies for ICC are underway (Table 3).
Novel methods for biomarker detection are also currently under investigation such as liquid biopsies. Liquid biopsies involve the detection of markers in patient fluid samples (e.g., blood, urine, or bile) that can be used to evaluate disease biology. Furthermore, liquid biopsies allow for serial detection of these markers that can provide information on changes in tumor biology [143]. Circulating tumor DNA (ctDNA) or RNA, circulating tumor cells (CTC), and tumor-derived exosomes, cytokines, and proteins are all biomarkers of interest. ctDNA may be used to assess response to systemic therapy. Ettrich et al. [144] investigated tumor tissue and corresponding ctDNA samples collected from patients with CCA prior to and during chemotherapy. Of note, blood and tissue were concordant in 92% of ICCs, and variant allele frequency in ctDNA correlated with tumor load and progression-free survival (PFS). Yang et al. [145] studied the use of CTCs in 88 patients with CCA and reported that 15 patients were positive for CTC; CTCs were associated with the extent of disease, more aggressive tumors, and predicted survival. Additionally, Han et al. [146] investigated the use of circulating microRNA as a bile-derived biomarker in cholangiocarcinoma and noted that microRNA represented the oncogenic characteristics of CCA tissue.
Identification of extracellular vesicles (EV) may play a role as an emerging biomarker for CCA. EVs are endocytic-oriented membrane vesicles released by tumor cells and are vital to regulate cellular microenvironments by transporting biologic material [147]. EVs may be involved in tumor-induced inflammation and chemoresistance [148,149]. Xu et al. [150] identified CCA-associated circRNA, circ-CCAC1, upregulated in bile EVs and tissues. circ-CCAC1 may serve as a biomarker or therapeutic target for cholangiocarcinoma. Likewise, the role of cir-cRNA in ICC is limited, but trials are ongoing [151]. Expression of circSMARC5 was decreased in ICC tissue and negatively correlated with advanced stage [152]. Similarly, circACTN4 promotes tumor cell growth by regulating Wnt signaling pathways and thus promotes tumor cell growth in ICC [153]. Therapeutic strategies could be developed to reduce the pro-oncogenic activity of circRNA. For example, target site blockers (TSBs) could target the miRNA response elements carried by circRNAs [151,154].

Conclusions
Patients with ICC have a poor prognosis despite multimodality therapy including resection with curative intent and systemic therapy. Furthermore, these therapies come with morbidity and toxicity for many patients. Emerging biomarkers may provide diagnostic utility and assist with treatment decisions for patients with ICC including appropriate patient selection for surgical intervention and individualized perioperative systemic therapy regimens. Furthermore, innovative investigative techniques, such as next generation sequencing, are expected to expand our knowledge of tumor biology and the underlying genetic and epigenetic drivers of disease. As a result, novel biomarkers will play an increasingly significant role in the management of patients with ICC. Future studies are required to evaluate novel biomarkers, as well as further define how to apply biomarkers in the clinical setting.

Author contributions
ZJB, DBH and TMP-Study Design; ZJB, DBH and TMP-Preparing the Manuscript; ZJB, DBH and TMPcritical review and revision of the manuscript.

Ethics approval and consent to participate
Not applicable.