Circulating tumor cells and cholangiocarcinoma

Potential conflict of interest: Nothing to report. Supported by a grant from the University of Milano‐Bicocca (project 2014‐ATE‐0287, to R.G.) and by the American Association for the Study of Liver Diseases Foundation's Alan Hofmann Clinical and Translational Research Award (to A.V.). See Article on Page 148 Cholangiocarcinoma (CCA) is the second most common primary hepatobiliary malignancy, with solid epidemiologic data suggesting that its incidence has steadily increased in the last two decades. CCA is generally classified according to its anatomical location along the biliary tree into intrahepatic, perihilar, and distal.1 Histological, clinical, and molecular data indicate that in addition to their different anatomical locations there may be inherited differences in terms of biological behavior among these CCA subtypes. Indeed, their clinical course and invasive behavior are quite different. Unfortunately, most patients are still diagnosed at advanced stages when palliative treatments provide limited survival benefit.1 In terms of molecular prognosis prediction, a recent study showed how nuclear expression of S100A4 in neoplastic ducts was associated with poor response to surgical therapy in CCA patients.2 In terms of circulating markers, there have been some refinements including the diagnostic role of CYFRA 21‐1 (a fragment of cytokeratin 19), which has been suggested to correlate with circulating tumor cells in biliary tract cancers.3 Recent reports using deep‐sequencing technologies identified candidate oncogenic drivers in intrahepatic CCA. In particular, gene fusions involving FGFR24 have been shown to induce tumor formation in experimental models of intrahepatic CCA, which highlights fibroblast growth factor inhibition as a potential new therapeutic strategy in these patients. Other common genetic defects include mutations in IDH and KRAS. Similar to hepatocellular carcinoma, clinical practice guidelines for CCA do not include any molecular data from the tumor in their recommended management strategy. The study of circulating tumor cells (CTCs) has emerged as a potential source of novel biomarkers in oncology. CTCs are cancer‐derived cells released from solid tumors into the bloodstream; they bear tumor‐initiation properties and can eventually drive metastasis formation, particularly when they group in clusters.5 In this context, the cell junction component plakoglobin seems to participate in CTC cluster formation and stabilization, being a potential new target to prevent metastatic cancer spread. A number of studies have evaluated the role of CTCs as biomarkers to predict recurrence, survival, or response to therapy in different tumor types. Most of them use CTC enumeration as a surrogate of metastatic potential and hence a marker of poor outcome. CTCs are relatively rare in the bloodstream and more likely to be detected in patients with metastatic disease, which puts into question its role as a biomarker at early tumor stages. The majority of CTC isolation methods rely on capturing cells that express cytokeratin and epithelial cell adhesion molecule (EpCAM) and that are also CD45‐negative so that immune cells are discarded. Other enrichment methods use physical properties of malignant cells such as size. The prognostic performance of CTCs in other liver tumors such as hepatocellular carcinoma has been thoroughly explored, mainly in studies using EpCAM‐based capture methods. Data from gene expression studies suggest that only a small fraction (10%‐20%) of hepatocellular carcinomas express EpCAM, so it is feasible that CTC enumeration may be underestimated in these studies. Interestingly, some of these reports even analyzed specific genetic traits within these CTCs, to ascertain their potential as drivers of metastasis formation. Data on the prognostic role of CTCs in CCA are scarce. In this issue of Hepatology, Yang et al.6 report an association of the number of CTCs with more aggressive tumor characteristics and lower survival in CCA patients. Although this is not the first analysis of CTCs in CCA,7 it is the largest of this kind performed so far that has included correlation with clinical outcomes. The study included CTC enumeration in a cohort of 88 patients using the semiautomated CellSearch (Janssen Diagnostics), the first CTC enumeration system to achieve US Food and Drug Administration approval. The study population was markedly heterogeneous, which somehow limits extrapolation of the outcome results. The authors found an association between CTC number and baseline tumor features suggestive of increased tumor burden such as size, multinodularity, and locoregional lymph node invasion. It was also correlated with overall survival using two different cutoffs, CTCs ≥2 and CTCs ≥5; and it was an independent predictor of survival in this data set along with other well‐known clinical variables. CTC enumeration allowed stratifying patients with extrahepatic disease because those with higher CTC numbers also have shorter survival. Similarly, subgroup analyses according to stage (i.e., American Joint Committee on Cancer/International Union Against Cancer tumor–node–metastasis system) confirmed the prognostic role of CTCs across stages, particularly when using the cutoff of CTCs ≥5. Unfortunately, the number of patients is relatively low in these subanalyses, so these results should be considered with caution. Also, the optimal CTC cutoff value has been under some debate in the oncology field, but solid data suggest that healthy individuals and those with nonmalignant conditions rarely have more than two circulating epithelial cells. There are not many data on CTCs in patients with primary sclerosing cholangitis, a well‐known risk factor for CCA. None of the primary sclerosing cholangitis‐CCA patients in the study by Yang et al.6 had detectable CTCs, but the numbers are still too small to derive a definitive conclusion on the presence of circulating epithelial cells as a potential confounding factor in patients with primary sclerosing cholangitis‐CCA. The study has some limitations. First one relates to the capture method, which relies on EpCAM expression and could eventually miss non‐EpCAM CTCs, so the actual CTC number could be underestimated. Paired analysis of EpCAM expression in tumor tissue with CTC enumeration could help us to understand the extent of this limitation. Also, the patient population included is quite heterogeneous, ranging from patients at early stages treated with surgery to those with metastatic disease who received chemotherapy. Along with this, some of the proposed subanalyses might be underpowered to accurately capture prognostic differences. Despite this, the study provides initial evidence of the prognostic role of CTCs in CCA and further reinforces the notion that circulating tumor biomarkers could have a role in decision making for patients with liver malignancies. CTC enumeration is framed within the concept of “liquid biopsy”, which encompasses the analysis of tumor by‐products in peripheral blood as a source of novel prognostic and predictive biomarkers.8 In addition to CTCs, the predictive value of circulating cell‐free DNA and RNA is being thoroughly evaluated in different malignancies. Of note, a recent report that compared the performance of both circulating tumor DNA and CTCs in detecting tumor‐specific molecular events suggests that the level of circulating tumor DNA tends to be higher than that of CTCs.9 It has also been suggested that tumor cells may actively release cell‐free nucleic acids to promote metastasis formation. For instance, exosomes released by pancreatic cancer cells induce liver premetastatic niche formation and increase metastatic burden.10 Exosomes are small membrane vesicles capable of transferring DNA, messenger RNA, and proteins between cells. High expression of macrophage migration inhibitory factor within pancreatic cancer cell–derived exosomes activates fibrotic and inflammatory pathways through Kupffer cell uptake, which constitutes a favorable milieu for metastatic deposit. Ultimately, a combined vector integrating molecular information from plasmatic nucleic acids coupled with data from CTCs may maximize the prognostic and predictive performance of liquid biopsy. The potential applications are endless, including patient stratification and evaluation of molecular tumor response after palliative treatments. As with any other potential new biomarker, confirmatory data generated on well‐designed studies will be critical; but certainly, liquid biopsy could become a very useful tool in the clinical management of patients with CCA. Author names in bold denote shared co‐first authorship.