Liquid biopsy in pancreatic ductal adenocarcinoma: current status of circulating tumor cells and circulating tumor DNA.

abstract cc by

Reliable biomarkers are required to evaluate and manage pancreatic ductal adenocarcinoma. Circulating tumor cells and circulating tumor DNA are shed into blood and can be relatively easily obtained from minimally invasive liquid biopsies for serial assays and characterization, thereby providing a unique potential for early diagnosis, forecasting disease prognosis, and monitoring of therapeutic response. In this review, we provide an overview of current technologies used to detect circulating tumor cells and circulating tumor DNA and describe recent advances regarding the multiple clinical applications of liquid biopsy in pancreatic ductal adenocarcinoma.

pancreatic ductal adenocarcinoma cc by

Pancreatic cancer is the fourth leading cause of cancer mortality in the United States (Kamisawa et al., 2016). In 2018, the American Cancer Society estimated that there will be 55 440 newly diagnosed cases and 44 330 deaths from pancreatic cancer (Siegel et al., 2018). Approximately 95% of pancreatic cancers are classified as exocrine cancers, while less than 5% of pancreatic cancers are endocrine cancers, namely, pancreatic neuroendocrine tumors. The exocrine cancers include pancreatic adenocarcinoma, acinar cell carcinoma, cystadenocarcinoma, and pancreatoblastoma: Pancreatic adenocarcinoma, or pancreatic ductal adenocarcinoma (PDAC), is the major histological subtype that comprises about 90% of all pancreatic cancers (Goel and Sun, 2015). The TNM stages of pancreatic cancer are based on American Joint Committee on Cancer (AJCC) Cancer Staging Manual, which consider primary tumor size (T), regional lymph node involvement (N), and distant metastasis (M) (Allen et al., 2017; Chun et al., 2018; Kamarajah et al., 2017; Kamisawa et al., 2016). Stages I and II are mostly considered as resectable, and stages III and IV are typically classified as locally advanced and metastatic, respectively. PDACs generally carry a very poor prognosis with the 5‐year survival rate for all stages of PDAC as low as 6–8% (Siegel et al., 2018; Ying et al., 2016). While surgical resection remains the only curative therapy, less than 20% of patients are candidates for surgical resection, which increases the 5‐year survival rate to 15–25% (Luketina et al., 2015; Schlitter et al., 2017). Approximately 50–60% of patients are found to have metastasis at diagnosis due to nonspecific or even lack of symptoms that limits earlier diagnosis (Kleeff et al., 2016), with only a 3% 5‐year survival for distant disease (Siegel et al., 2018). Clinicians are struggling to develop diagnostic strategies for the early detection of the disease. Adequate biopsy is still challenging because of its poor anatomic location. Endoscopic ultrasound‐guided fine needle aspiration (EUS‐FNA) is preferred for obtaining specimens for biopsy, yet its negative predictive value remains at 16–86% (Mohammad Alizadeh et al., 2016). Currently, the serum level of CA 19‐9 is a widely used biomarker for the diagnosis or monitoring of PDAC, but CA19‐9 alone exhibits a wide range of sensitivity (70–95%) and specificity (70–90%) (Ballehaninna and Chamberlain, 2012; Scara et al., 2015). False‐negative results are observed in patients with the Lewis‐negative blood group, Le(a‐b‐) that occurring in about 5–10% of Caucasians, and false‐positive results have been reported in other diseases including obstructive jaundice, acute cholangitis, and chronic pancreatitis (Passerini et al., 2012; Tanaka et al., 2000). Thus, a highly sensitive, reliable, and noninvasive biomarker for evaluating and managing PDAC patients is still required.

circulating tumor cells and circulating tumor dna cc by

Circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA), as liquid biopsies, are an emerging minimally invasive tool for cancer diagnosis, surveillance, and treatment. CTCs can be traced back to their first description by Thomas Ashworth in 1869 (Ashworth, 1869). CTCs are released from primary tumor and/or metastatic sites into the bloodstream. Since CTCs exist as rare cells in the blood (one CTC among 106–109 blood cells), recent studies focus on the efficient capture of rare CTCs from whole blood (Ferreira et al., 2016). Investigators use CTCs as a guide to (a) determine prognosis, (b) monitor in real‐time therapeutic responses and tumor recurrence, (c) explore therapeutic targets, and (d) potentially develop new drugs by studying metastatic cancer biology and drug resistance mechanisms in CTCs (Ferreira et al., 2016). ctDNA is a subset of circulating extracellular DNA in plasma (also called cell‐free DNA, cfDNA), specifically released from cancer cells. ctDNA (known as tumor‐derived cfDNA) may originate from apoptotic and necrotic tumor cells, from living tumor cells, or even from CTCs; thus, it has a variable half‐life from 15 minutes up to 2 h (Alix‐Panabieres and Pantel, 2016; Diaz and Bardelli, 2014; Diehl et al., 2008; Kidess and Jeffrey, 2013; Nordgard et al., 2018). While the size of cfDNA released by apoptotic cells represents approximately 166 bp, ctDNA has recently been reported as being more highly fragmented (Mouliere and Rosenfeld, 2015; Underhill et al., 2016). Detecting ctDNA is generally based on the target mutation (e.g., KRAS, BRAF, EGFR, hypermethylation, and multiple gene panels) (Kidess and Jeffrey, 2013). Due to its small fraction (occasionally < 0.01%) among total cfDNA in circulation, approach with sensitive detection methods for ctDNA is highly recommended (Cheng et al., 2016). Recent advances in ctDNA analysis highlight future critical roles in cancer management of this easily and serially accessible assay: (a) monitoring tumor burden, (b) evaluating therapeutic response, and (c) identifying therapeutic targets through minimally invasive molecular profiling (Ignatiadis et al., 2015). Intratumoral heterogeneity exists due to uneven distribution of cancer subclones in the same tumor (spatial heterogeneity) and due to different genetic alterations that may be selected over time (temporal heterogeneity) as a result of microenvironmental selection, genomic instability, and following multiple drug treatments, where such treatments would ablate cancer cells sensitive to the treatment but not block expansion of residual surviving drug‐resistant cancer cell subpopulations (Dagogo‐Jack and Shaw, 2018; Friedman, 2016; Jeffrey and Toner, 2019; McGranahan and Swanton, 2017). Moreover, most patients with metastatic cancer have multiple rather than solitary metastases, some of which may be discordant with the primary tumor and between other metastases. Sequential tissue sampling of every metastatic lesion is impractical and risky. As a liquid biopsy represents cancer cells or cancer cell products/nucleic acids derived from the entire tumor burdens of the patient, liquid biopsy can be a valuable alternative to tissue biopsies. The following discussion summarizes the current technologies of CTCs and ctDNA and application of these tools to manage patients with PDAC.

current technologies in ctcs cc by

Current CTC technologies include two main steps: CTC enrichment and CTC identification. CTC enrichment strategies focus on improving yield of capturing tumor cells, called capture efficiency, and obtaining high‐purity CTCs via depleting the background blood cells (i.e., leukocytes). The most widely used enrichment strategies are based on immunoaffinity, called label dependent, which uses cell surface markers to capture epithelial tumor cells. Immunomagnetic capture is widely used: The specific antibodies are normally conjugated with magnetic nanoparticles, and a magnetic field is then used to capture the CTCs. Tumor‐specific cell surface antigens, such as epithelial cell adhesion molecule (EpCAM), are targeted for the positive enrichment: CellSearch®, which is the only US Food and Drug Administration‐approved platform, MACS®, and MagSweeper are examples that may use EpCAM‐based or other markers, while AdnaTest uses a cocktail of antibodies against multiple antigens (e.g., EpCAM, EGFR, and HER2). In contrast, negative enrichment is the depletion of nonspecific background cells (i.e., leukocytes) using anti‐CD45 antibodies not expressed by tumor cells: MACS®, Quadrupole Magnetic Sorter (QMS), Dynabeads®, and EasySep™ are based on this strategy. Antibodies can also be attached to microposts and other surfaces for CTC capture. Microfluidic devices have been developed based on the technology controlling the fluid flow, which offers advantages for CTC research such as improved capture efficiency and high purity (Warkiani et al., 2016). The geometrically enhanced differential immunocapture (GEDI) device uses geometrically enhanced microstructures and combines positive enrichment with hydrodynamic chromatography, which additionally enables cell size‐based separation. Surface‐capture microfluidic devices, such as Herringbone (HB) Chip, Geometrically enhanced mixing (GEM) chip, Graphene oxide (GO) Chip, and the modular sinusoidal system (BioFluidica), increase collision events between the cells and the surface‐coated antibodies. The other kind of microfluidic devices, such as CTC‐iChip, IsoFlux™, LiquidBiopsy, Ephesia chip, and Magnetic Sifter, use microfluidic‐ and immunomagnetic‐based strategies, and these devices exhibited higher sensitivity in CTC separation than CellSearch® (Karabacak et al., 2014). Another major type of CTC enrichment strategies, known as label‐independent enrichments, relies on biophysical properties (e.g., size, including inertial focusing, electrical charge, and density). A substantial number of microfiltration systems are based on the principle that tumor cells (12–25 μm) are basically larger than leukocytes (8–14 μm) (Sollier et al., 2014). Therefore, these systems use 7–8 μm pores [isolation by size of epithelial tumor cells (ISET) filter device, ScreenCell®, and CellSieve™], or less (VyCAP microsieves which have a membrane thickness smaller than the pore size), microfabricated filter membranes [Flexible Micro Spring Array (FMSA) (Harouaka et al., 2014)], or 3‐dimensional microfiltration layers (FaCTChecker, Resettable Cell Trap, and Cluster‐Chip). Inertial focusing microfluidics can be applied for size‐based separations (Vortex and ClearCell® systems). Dielectrophoresis (DEP) uses the polarizabilities of cells in a nonuniform electrical field. In the electrical field, cells are pushed by either negative or positive force and separated based on their cell size and polarizability. Commercialized DEP systems include ApoStream® and DEPArray™. Recently, microfluidic platforms applied both cell size‐ and deformability‐based systems for CTC enrichment: The Parsortix™ (Xu et al., 2015, 2017) and Celsee™ (Gogoi et al., 2016). A density‐based gradient technology has been also commercialized for separating CTCs: Ficoll‐Paque®, RosetteSep™, OncoQuick®, and Lymphoprep™. Viable CTCs can be further characterized through combining functional assay with capturing CTCs (Alix‐Panabieres et al., 2016). The method that targets secreted tumor‐associated analytes [i.e., Epithelial ImmunoSPOT Assay (EPISPOT)] and the assay based on cell adhesion matrix (CAM) (i.e., Vita‐Assay™ and Vita‐Cap™) are commercially available (references for technology platforms described above are cited in (Ferreira et al., 2016). After enrichment of CTCs, verification of the captured cells is subsequently required. Immunofluorescence (IF) staining, which usually defines 4′,6‐diamidino‐2‐phenylindole (DAPI) + (nuclear stain), CD45− (leukocyte marker), and cytokeratin (CK) + (epithelial marker), which identify epithelial‐like CTCs, is most extensively used, but immunohistochemical (IHC) staining using chromogenic reporters, fluorescence in situ hybridization (FISH), and molecular analyses ranging from reverse transcription polymerase chain reaction (RT–PCR) to aptamer‐based assays to targeted sequencing is also used (Paterlini‐Brechot and Benali, 2007; Smith et al., 2007; Swennenhuis et al., 2009).

clinical application of ctcs in pdac cc by

Previous CTC studies in pancreatic cancer are summarized in Table 1.

detection cc by

The detection of CTCs in patients of pancreatic cancer has been compared with that in patients with other cancers in previous studies. Using the CellSearch® system, Allard et al. enumerated CTCs in 2183 blood samples from 946 metastatic patients with 12 different cancer types, which included 21 blood samples from 16 patients with pancreatic cancer. Lower number of CTCs was detected in pancreatic cancer (mean, 2 CTCs/7.5 mL) than any other carcinomas, such as prostate cancer, ovarian cancer, breast cancer, gastric cancer, colorectal cancer, bladder cancer, rental cancer, and lung cancer. CTCs above the cutoff level (≥2 CTCs) were detected in only 4 out of 21 samples (19%) (Allard et al., 2004). In contrast, recent works using state‐of‐the‐art techniques demonstrated comparable detection rates of CTCs in pancreatic cancer when compared with those in different types of carcinomas. Zhang et al. (2016a) used hTERT promoter‐regulated oncolytic herpes simplex virus‐1 that targets telomerase reverse transcriptase‐positive tumor cells, and identified CTCs in 88.2% (15/17) of patients with various stages of pancreatic cancer. Chang et al. developed a parallel flow microfluidic chip that is combined with different strategies such as immunomagnetics and size‐based filtration. This device performed well for isolating of CTCs in patients with metastatic pancreatic cancer (91.7%, 11/12 in pancreatic cancer; 100%, 38/38 in non‐small‐cell lung cancer) (Chang et al., 2015). Another study by Ting et al. applied the microfluidic CTC‐iChip, which depletes normal blood cells by inertial focusing size‐based sorting and separates CTCs immunomagnetically, for single‐cell RNA sequencing. In this study, median 118 CTCs/mL (range, 0–1694) were detected in pancreatic tumor‐bearing mice (KPC mice) (Ting et al., 2014). Varillas et al. (2017) have introduced a detailed procedure for using a microfluidic chip with a herringbone structure and reported that this device could consistently detect a low number of CTCs in pancreatic cancer. Interestingly, El‐Heliebi et al. applied KRAS as a marker for CTC enumeration and molecular characterization. They used an in vivo isolation of CTCs (GILUPI CellCollector®) directly from the vein of patients and applied signal amplification of in situ padlock probes via rolling‐circle amplification: 47% (7/15) of patients were CTC‐positive (range, 1–3 CTCs/patient), and 40% (6/15) of patients had KRAS mutant CTCs (El‐Heliebi et al., 2018). With regard to the enrichment strategies, size‐based filtering strategies exhibited higher sensitivity in isolating CTCs compared with EpCAM‐based approaches in patients with metastatic or inoperable pancreatic cancer: ISET and CellSearch® detected CTCs in 88.9% (38/50) and in 39.6% (21/53) of patients, respectively (Khoja et al., 2012). A recent study by Brychta et al. compared the performance of these two strategies by cell spiking experiments [EpCAM‐based CTC isolation (IsoFlux) vs. automated size‐based filtration (Siemens Healthineers)]: Especially for low EpCAM expressing cells, the filtration‐based strategy exhibited higher recovery rate (52%) than the IsoFlux device (1%). Additional experiments using the filtration‐based strategy were able to capture CTCs in 42% of frozen diagnostic leukapheresis (DLA) samples from 19 patients with pancreatic cancer. Although there was no difference in prevalence of CTCs in samples from patients with and without metastases (44% vs 40%, respectively), CTC numbers were somewhat higher when distant metastases were present (0–7 for Stage IV disease versus 0–2 for stages 2b‐III) (Brychta et al., 2017).

early diagnosis cc by

The potential role of CTCs as an early diagnostic marker has recently been revealed by Rhim et al. (Table 2). Using GEDI chip, CTCs were captured in three different subject groups [PDAC patients at all stages, patients with precancerous cystic lesion, that is, intraductal papillary mucinous neoplasm (IPMN) or mucinous cystic neoplasm, and cancer‐free controls]. Interestingly, CTCs were detected in 40% (8/21) of the patients with precancerous lesions: Circulating pancreas epithelial cells may precede the detectable tumors. The detection rates of CTCs were 73% (8/11) and 0% (0/19) in PDAC patients and cancer‐free group, respectively (Rhim et al., 2014).

a marker of advanced disease cc by

The correlation of CTC levels with more aggressive pathologic features and with advanced disease is still debated. A multicenter randomized clinical trial suggested that CTC detection with CellSearch® significantly correlated with aggressive tumor differentiation (Bidard et al., 2013). In another study, which used a modular microfluidic system, CTC levels isolated from metastatic PDAC patients (mean 53 CTCs/mL, n = 7 patients) was significantly higher than those from resectable PDAC patients (mean 11 CTCs/mL, n = 5 patients), although further testing will be required because of the small numbers of patients tested in this first proof‐of‐principle assay (Kamande et al., 2013). The expression of C‐MET, CK20, and CEA mRNA detected by RT–PCR after MACS purification correlated with TNM stage (Zhou et al., 2011). More recently, Court et al. (2018) preoperatively enumerated CTC using the microfluidic NanoVelcro chip and reported that PDAC patients with occult metastatic disease had significantly more CTCs than PDAC patients with localized disease (median 7 CTCs vs. 1 CTC, P < 0.0001). In contrast, Cauley et al. (2015) described that CTC positivity was not associated with tumor characteristics, lymph node metastasis, respectability, and advanced TNM stage. Similarly, the percentage of CTC detection using size‐based filtration was not associated with the TNM stage or distant metastasis (Bobek et al., 2014; Kulemann et al., 2015).

prognosis cc by

Studies investigating the role of CTC detection as a prognostic marker are summarized in Table 3. Research efforts on CTC enumeration for better prognostic classification are well underway. Several studies discussed below performed multivariable analysis using the Cox regression model, which exhibits CTCs as an independent prognostic factor. Bidard et al. (2013) conducted multicenter randomized clinical trial evaluating 79 patients with locally advanced nonmetastatic PDAC. Patients were randomly assigned to receive gemcitabine alone, or gemcitabine plus erlotinib. The CTC positivity was measured by CellSearch® at two different time points (at baseline and at two months): The overall detection rate of CTCs (either at baseline or at two months) was 11%. CTC positivity in locally advanced pancreatic adenocarcinoma at any time point was an independent prognostic factor for overall survival (OS) in multivariable analysis but not for progression‐free survival (PFS). A more recent study by Effenberger et al. enrolled 69 patients with PDAC and identified CTCs using MACS enrichment: Here, CTC positivity was an independent risk factor of reduced PFS (HR = 4.543, P = 0.006) and OS (HR = 2.093, P = 0.028) (Effenberger et al., 2018). Studies using different platforms in PDAC patients exhibited association of CTCs with survival rates. Chang et al. used a supported lipid bilayer (SLB) surface‐coated microfluidic chip (CMx platform): Patients with unfavorable circulating tumor microemboli (CTM) levels exhibited shorter PFS and OS when compared with patients with favorable CTM levels (PFS, 2.7 months vs. 12.1 months, P < 0.0001; OS, 6.4 months vs. 19.8 months, P < 0.0001). These associations were still observed in each subgroup (early stage and advanced stage) (Chang et al., 2016). Gao et al. (2016) applied EpCAM independent subtraction enrichment and immunostaining‐FISH (SE‐iFISH) to enumerate CTCs and demonstrated that the presence of ≥ 3 CTCs/7.5 mL was the strong predictive factor for worse OS (HR = 4.547, P = 0.016). Poruk et al. compared epithelial CTCs and mesenchymal‐like CTCs using IF staining for panCK and vimentin markers, respectively, after the size‐based CTC separation. The epithelial CTCs (CK‐positive) were strongly associated with poorer survival but not mesenchymal‐like CTCs (P < 0.01 vs. P = 0.39). With regard to median time to recurrence, detection of CTCs expressing both CK and vimentin was the significant predictive factor for earlier recurrence (P = 0.01) (Poruk et al., 2016). A recent useful meta‐analysis described that detectable baseline CTCs including disseminated tumor cells in the bone marrow was associated with worse disease‐free survival (DFS)/PFS (HR = 1.93, P = 0.007) and OS in pancreatic cancer (HR = 1.84, P ≤ 0.0001) (Stephenson et al., 2017).

different sampling sites cc by

Research comparing CTCs in portal vein (PV) and those in peripheral blood (PB) is in progress (Table 4). Bissolati et al. evaluated PV samplings in 20 patients with nonmetastatic PDAC undergoing surgical resection. Five out of nine CTC‐positive patients had CTCs in PV but not in systemic circulation, detected by CellSearch®. At 3‐year follow‐up, patients with detectable CTCs in PV exhibited higher rate of liver metastasis than patients without detectable CTCs in PV (53% vs. 8%, P = 0.038) (Bissolati et al., 2015). Catenacci et al. evaluated CTCs in EUS‐guided PV sampling. Using CellSearch®, they detected CTCs in PV blood samples from 100% (18/18) of patients, while only four patients (22.2%) had CTCs in the PB. Even in patients with nonmetastatic and localized or borderline‐resectable pancreatic cancer, high levels of CTCs were detected (mean 83.2 CTCs/7.5 mL) in PV (Catenacci et al., 2015). Further recently, Tien et al. (2016) collected intraoperative PB and PV samples from 41 PDAC patients. CTC count (CMx platform) in PV was a strong predictor for liver metastasis in a 6‐month follow‐up after surgery (P = 0.002). The PV is the main entrance for distant metastasis of PDAC, and tumor cells spread into blood circulation before radiologically detected. CTCs in PV seem to more closely reflect the metastatic potential, although prospective studies with large cohorts are still required.

additional markers for ctcs in pdac cc by

Epithelial–mesenchymal transition (EMT) may explain how the epithelial tumor cells disseminate from primary site and penetrate the endothelium of blood vessel (Chaffer and Weinberg, 2011). Even though the extent of tumor cells undergoing EMT still remains unclear, the epithelial markers (e.g., EpCAM and CK) of epithelial cells are downregulated by EMT‐inducing signals; thus, CTC capture strategies targeting expression of epithelial markers may fail to isolate a subset of CTCs (Krebs et al., 2014). The expression of epithelial markers such as EpCAM, CK, and E‐cadherin has been reported to be reduced lower than 40% in CTCs of PDAC (Rhim et al., 2012). Similarly, CellSearch® detected CTCs in 39.6% (21/53) of patients with metastatic PDAC, while ISET exhibited better enrichment of CTCs (CTC positivity in 88.9% of patients with metastatic PDAC) (Khoja et al., 2012). Combining additional markers for capturing mesenchymal‐like CTCs remain to be identified. Potential mesenchymal markers include the following: ZEB1, SNAI1, vimentin, N‐cadherin, FGFR2, PLS3, Twist1, and PI3K/AKT (Barriere et al., 2014). A few recent studies have reported the application of mesenchymal markers to detect CTCs in PDAC. CTCs enriched by ScreenCell® filtration devices were stained with ZEB1 and CK. ZEB1‐positive CTCs were found in almost exclusively in patients with metastatic PDAC (P = 0.01) (Kulemann et al., 2016). Dotan et al. evaluated 23 patients with metastasis who had at least one CTC detected at baseline by using CellSearch®. They assessed for the expression of MUC‐1, which play a role of inducing EMT: MUC‐1 expression was observed in 43% (10/23) of the patients, and patients with CTCs positive for MUC‐1 had shorter median OS than those with CTCs negative for MUC‐1 (2.7 months vs. 9.6 months, P = 0.044) (Dotan et al., 2016). Another study, which compared epithelial CTCs and mesenchymal‐like CTCs using a vimentin marker, was discussed above (Poruk et al., 2016). However, blood cells including monocytes and granulocytes retain vimentin expression during the maturation, which warrant additional confirmation of tumor‐specific markers (Dellagi et al., 1983). A subset of tumor cells, so‐called cancer stem cells (CSCs), have properties of stem cells and display self‐renewing and multipotency capabilities, which are considered to be responsible for metastasis, chemoresistance, and recurrence of tumors (Krebs et al., 2014; Satoh et al., 2015). It has been reported that CSC and EMT share common molecular pathways (e.g., Wnt/ß‐catenin and Notch signaling), and epithelial cells undergoing EMT acquire CSC features (Igawa et al., 2014). Key markers for identifying pancreatic CSCs include CD133 and aldehyde dehydrogenase (ALDH) (Fitzgerald and McCubrey, 2014). Marker combinations of CD44, CD24, and epithelial‐specific antigen (ESA) were also identified as indicators of pancreatic CSCs (Li et al., 2007). Other putative markers for pancreatic CSCs include c‐Met, doublecortin‐like kinase 1, and CD44v6 (Polireddy and Chen, 2016). A recent study by Poruk et al. evaluated 60 consecutive PDAC patients undergoing surgery. CTCs were detected by IF staining using CK, CD133, CD44, and ALDH, after isolated by ISET. CK+/ALDH+ CTCs and CK+/CD133 + /CD44 + CTCs were detected in 77% (46/60) and in 57% (46/60) of patients, respectively. For the 59 nonmetastatic patients, ALDH‐positive CTCs and CK+/CD133 + /CD44 + CTCs were significantly associated with decreased DFS and higher risk of tumor recurrence (Poruk et al., 2017).

current technologies in ctdna cc by

Since ctDNA is present in minute quantity in the bloodstream, extraction of cfDNA without contamination of plasma with genomic DNA is a major challenge in ctDNA analysis. Preanalytical variables that include specimen types (plasma or serum), specimen collection procedures (time to processing of whole blood), blood collection tubes, specimen handling (including centrifugation protocols and temperature), and methods of cfDNA isolation and purification are the most important factors to control this success (Diefenbach et al., 2018; Markus et al., 2018; Sato et al., 2018). Plasma has been preferred as a source for extracting circulating DNA. Even though serum contains 2–24 times higher amount of cfDNA than plasma, serum is not recommended due to the possible contamination from white blood cells during the clotting process (Heitzer et al., 2015; Parpart‐Li et al., 2017; Trigg et al., 2018; Zhao et al., 2019). If specimen processing can be performed within 6 h from collection, standard K2EDTA collection tubes are suitable for blood sampling. However, when the processing is delayed by up to 48 h, specialized cell‐stabilizing blood collection tubes should be used to reduce contamination by genomic DNA released from leukocyte lysis (Alidousty et al., 2017; Medina Diaz et al., 2016; Merker et al., 2018; Risberg et al., 2018; Ward Gahlawat et al., 2019; Warton et al., 2017). Current evidence recommends that isolated plasma, not whole blood, can be stored frozen up to 9 months or up to a few years, depending on analytical goals (van Dessel et al., 2017; Meddeb et al., 2019). The isolated plasma is preferably aliquoted into a single use fraction: A single freeze–thaw cycle had no significant effect on cfDNA stability (Bronkhorst et al., 2015; Merker et al., 2018). Several issues regarding DNA isolation and nonmalignant conditions that induce the release of cfDNA should be considered, but the following discussion focuses more on the techniques in progress for sensitive detection of the small fraction of ctDNA (Heitzer et al., 2015; Qin et al., 2016). Based on PCR technology, new technologies including real‐time quantitative PCR (qPCR) (Brown, 2016), amplification‐refractory mutation system (ARMS)‐based qPCR (Zhang et al., 2015c), competitive allele‐specific TaqMan PCR (cast‐PCR) (Ashida et al., 2016; Reid et al., 2015), coamplification at lower denaturation temperature PCR (COLD‐PCR) (Milbury et al., 2011) have been introduced. More recently, digital PCR (dPCR), which uses droplets to compartmentalize individual DNA strands, reached the high sensitivity ranging from 0.1% to 0.001% and is therefore beneficial to detect low allele frequency variants (Gorgannezhad et al., 2018; Vogelstein and Kinzler, 1999). dPCR includes droplet PCR, Bio‐Rad droplet dPCR (ddPCR) platform (Hindson et al., 2011), and BEAMing (beads, emulsion, amplification and magnetics) (Chen et al., 2013): This method is currently among the most promising of targeted approaches, which focuses on the detection of rare mutations in DNA samples with prior knowledge of genetic changes at specific loci of the tumor (e.g., KRAS, BRCA2, ERBB2, and EGFR) (Alix‐Panabieres and Pantel, 2016; Cheng et al., 2017) and exhibits high analytical sensitivity. BEAMing combines emulsion PCR amplification and flow cytometry and therefore can be assessed in the standard laboratory setting (Dressman et al., 2003). BEAMing quantifies independently the fluorescently labeled particles, which is able to detect the rare variants with allele frequency < 0.01%. This method enables the counting of error rate of DNA polymerases (Gorgannezhad et al., 2018). The ddPCR platform performs PCR amplification within water‐in‐oil emulsion droplets where individual DNA molecules are dispersed in. Using fluorescently labeled probes, droplets can be identified as a binary (mutant‐positive or mutant‐negative) system. The Bio‐Rad QX‐200 platform produces 20 000 droplets and is one of the most commonly used dPCR systems for ctDNA detection (Gorgannezhad et al., 2018). Next‐generation sequencing (NGS), or a massively parallel sequencing, detects a wider range of mutation with higher coverage, but with lower sensitivity (approximately 1%) than dPCR. The targeted NGS approach sequences multiple cancer‐associated genes (Zill et al., 2015). Platforms such as safe‐sequencing system (Safe‐SeqS) (Kinde et al., 2011), TAm‐Seq (Forshew et al., 2012), Ion‐AmpliSeq (Rothe et al., 2014), CAPP‐Seq (Newman et al., 2014), and sensitive mutation detection using sequencing (SiMSen‐seq) (Stahlberg et al., 2017) have been developed. Zill et al. used Guardant360 assay to sequence cfDNA in 21 867 advanced cancer patients including 867 PDAC samples and reported the genomic findings and the response outcomes (Zill et al., 2018). Recent progress enabled whole‐genome sequencing to be applied to a liquid biopsy (Dawson et al., 2013). These NGS approaches largely extended noninvasive profiling of tumors not only focus on single nucleotide variants but also identify structural variants and copy number variations [e.g., personalized analysis of rearranged ends (PARE)] (Leary et al., 2012). Recent advances in NGS technology enable similar sensitivity to detection of ctDNA as by digital PCR. A recent study showed a statistical method based on each base‐position error rate (BPER), which detects variants with low allele frequency as low as 0.003 (single nucleotide variation) and 0.001 (insertions/deletions) (Pécuchet et al., 2016). Newman et al. recently developed an integrated digital error suppression (iDES)‐enhanced CAPP‐Seq, which incorporates in silico removal of artifacts detected in cfDNA sequencing data. This strategy enabled very sensitive detection of tumor‐derived DNA down to 0.002% for generalized iDES‐enhanced CAPP‐Seq and 0.00025% using a customized panel (Newman et al., 2016). Other newer methods include the use of bar‐coded amplicon‐based NGS rather than hybrid capture‐based plasma NGS (Guibert et al., 2018) or an improved method using dual peptide nucleic acid (PNA) clamping‐mediated locked nucleic acid‐dual peptide nucleic acid PCR clamp (LNA‐dPNA PCR clamp) with sensitivities in the 0.01%‐0.1% range (Zhang et al., 2019). Figure 1 summarizes the various technologies and the rages of their limit of detection. There has been encouraging improvement in the quest for early detection of pancreatic cancer. CancerSEEK, a multi‐analyte blood test, combines multiplex PCR (16 genes) and immunoassay (8 protein biomarkers) (Cohen et al., 2018). This method has shown over 69% sensitivity and over 99% specificity for five cancers including pancreatic cancer aiming to screen different cancers in the general population (Kalinich and Haber, 2018).

clinical application of ctdna in pdac cc by

Previous ctDNA studies in pancreatic cancer are summarized in Table 5. Several studies demonstrated that the amounts of plasma DNA in patients with cancer is higher than those in healthy individuals (Anker et al., 1999; Sozzi et al., 2003). With regard to PDAC, multiple studies have reported that cfDNA concentration was higher in pancreatic cancer patients compared with normal controls and in advanced stages compared with early stages (Berger et al., 2016; Singh et al., 2015; Takai et al., 2015).

method comparison cc by

Pécuchet et al. evaluated 77 patients with pancreatic cancer and compared a microfluidic dPCR (RainDrop®) and NGS analysis (Ion Proton™) in detecting KRAS and EGFR mutations. 97.4% (75/77) of results were concordant. KRAS mutation was only detected by dPCR in two samples (allele frequency 0.003 and 0.006, respectively) (Pécuchet et al., 2016). Similarly, Pietrasz et al. assessed 135 patients with PDAC and compared the two methods in detecting KRAS mutant ctDNA. They reported high concordance (R 2 = 0.94) between the targeted NGS analysis (Ion Proton™) and dPCR (RainDrop®) in detecting KRAS mutant ctDNA: One sample considered as KRAS mutation‐negative in NGS analysis was positive in dPCR (allele frequency 0.0061) (Pietrasz et al., 2017). Takai et al. applied a two‐stage strategy to analyze KRAS mutant ctDNA in PDAC patients. They used ddPCR (Bio‐Rad) as a prescreening method and then performed NGS analysis (Illumina HiSeq 2000) for 60 genes including KRAS (Takai et al., 2015). The use of NGS analysis as a prescreening method, combined with ddPCR (Bio‐Rad) for further validation, has also been successfully applied for ctDNA analysis (Adamo et al., 2017; Cheng et al., 2017). The combined strategy was suggested as cost‐effective and efficient method for analyzing ctDNA in PDAC patients (Takai et al., 2015). Further approaches to establish efficient strategies for analyzing tumor genomes in plasma DNA are highly warranted.

early diagnosis cc by

According to the recent joint review by the American Society of Clinical Oncology (ASCO) and the College of American Pathologists (CAP), further studies are still required to prove the clinical utility of ctDNA in early diagnosis (Merker et al., 2018). IPMNs are the most frequent potentially malignant pancreatic cysts and classified into main duct type (MD‐IPMN) and branch duct type (BD‐IPMN). Since only 15–20% of BD‐IPMN will develop malignancy and nonsurgical management is recommended for low‐risk BD‐IPMNs, we need to correctly identify malignant IPMNs. Recently, an imaging tool that is combined with the identification of genomic patterns, coined ‘radiomics’, has been proposed by several studies (Hanania et al., 2016; Permuth et al., 2016). Similarly, Berger et al. detected GNAS mutant plasma DNA in 71.4% (15/21) of IPMN patients, but neither in serous cyst adenoma patients nor in healthy controls. ctDNA assay can be a useful tool for the discrimination of IPMN with malignant potential from other harmless pancreatic tumors, even though additional approaches to differentiate low from high‐grade IPMN is still required (Berger et al., 2016). For realizing early cancer detection using ctDNA‐based screening tests, an interesting clinical trial (https://clinicaltrials.gov/ct2/show/NCT02889978) by a company (GRAIL, Inc) is currently ongoing and recruiting 15 000 participants including cancer subjects with multiple types and healthy subjects. This project, called the Circulating Cell Free Genome Atlas (CCGA), aims to identify potential cancer mutations and to complete a reference database of the mutations in circulating DNA in plasma (Aravanis et al., 2017).

prognostic marker cc by

Previous ctDNA studies mostly focused on KRAS hotspot (codon 12) mutations and its association with clinical outcomes of patients with PDAC. Sausen et al. (2015) demonstrated that patients with KRAS mutant ctDNA after surgery were more likely to relapse than those without KRAS mutant ctDNA (9.9 months vs. not reached, P = 0.02). Another study evaluated PDAC patients undergoing surgery and reported that the detection of ctDNA by ddPCR at baseline correlated with shorter DFS and OS (DFS, 6.1 months vs. 16.1 months; OS, 13.6 months vs. 27.6 months; P < 0.001 and P < 0.0001, respectively) (Hadano et al., 2016). This was also confirmed by Earl et al. (2015) in which patients with ctDNA detected by ddPCR had significantly shorter OS than patients with no detectable ctDNA. In metastatic PDAC, undetectable KRAS mutant ctDNA was significantly associated with survival benefit (8 months vs. 37.5 months, P < 0.004) (Perets et al., 2018). For patients with resectable disease, MST of patients in whom ctDNA was detected were significantly shorter than those of patients in whom ctDNA was not detected (3.9 months vs. 10.2 months, P < 0.001) (Chen et al., 2010). Furthermore, it has been reported that high amount of cfDNA is a relevant prognostic marker for pancreatic cancer patients (Singh et al., 2015; Tjensvoll et al., 2016). A recent meta‐analysis by Creemers et al. (2017) showed that the ctDNA in pancreatic cancer is significantly associated with a poor prognosis. In contrast, Bernard et al. (2019) analyzed longitudinal KRAS mutant allele fraction from ctDNA and exosome DNA and determined that longitudinal monitoring through exosome DNA rather than ctDNA provides prognostic information.

predictive marker cc by

So far, the role of ctDNA as a relevant predictive marker in PDAC remains to be identified. Recently, reported predictive markers for gemcitabine response are limited to the germline variants (Innocenti et al., 2012; Li et al., 2016). In a phase III trial, comparing gemcitabine alone with erlotinib plus gemcitabine, the OS was significantly prolonged on the combined therapy, yet EGFR status did not predict the response to the therapy (Moore et al., 2007). As the frequency of KRAS mutation in PDAC ranges from 88 to 100%, current efforts are underway to target KRAS pathway to make therapeutic progress in PDAC (Collisson et al., 2012; Krantz and O'Reilly, 2018; Rao et al., 2004; Van Cutsem et al., 2004; Ying et al., 2016). Additionally, targeting pancreatic CSCs, γ‐secretase inhibitors (GSI) to inhibit Notch signaling pathway have been developed (Abel et al., 2014; Whitehead et al., 2012). With regard to the epigenetic regulation, deregulation of histone deacetylases (HDACs) has been reported to play a role in pancreatic cancer development (Polireddy and Chen, 2016). HDAC inhibitors are currently tested for pancreatic cancer treatment, but there seems to be no benefit in clinical outcomes (Millward et al., 2012; Richards et al., 2006; Tinari et al., 2012). In this context, ctDNA assay will have clinical utility in noninvasive molecular profiling for the novel druggable mutations. An NGS approach targeting 60 cancer‐associated genes identified potentially targetable mutations in plasma DNA of PDAC patients (Takai et al., 2015).

future perspectives cc by

Early detection, real‐time disease monitoring, molecular profiling for targeted therapy are applications that promise to improve pancreatic cancer management. Liquid biopsy is a potentially valuable tool for in this regard. Multiple studies revealed the clinical use of liquid biopsy in monitoring patients (Table 6). ctDNA analysis may be more sensitive, easily accessible, and suitable not only for monitoring tumor dynamics during treatment, but for noninvasive molecular profiling of tumors due to the high incidence of nongermline (as well as some germline) genetic variations (Cicenas et al., 2017). Several recent studies have performed ctDNA analysis targeting noncoding repetitive DNA sequence such as ALU and described the possible use of noncoding DNA as additional prognostic marker in cancer monitoring (Chang et al., 2017; Lehner et al., 2013). CTC analysis, however, has its own strengths in that CTCs enable functional analyses such as drug testing, particularly as they represent cells still remaining after previous treatment during the course of disease. Thus, we suggest that both CTCs and ctDNA can be used in future parallel or complementary analyses (Kidess‐Sigal et al., 2016) and it is hoped that both these technologies will influence future diagnosis and treatment of this currently devastating disease. In addition to CTCs and ctDNA, there is increasing attention for emerging role of extracellular vesicles (EVs). Exosomes are a well‐studied EV population and can be a source for tumor‐specific proteins and RNAs (i.e., mRNA, noncoding RNA, and miRNA). Exosomes that carry cargo consisting of disease‐specific nucleic acids and proteins can provide a promising tool for characterizing cancer specific features as well as targeted treatment in pancreatic cancer (Kamerkar et al., 2017; Massoumi et al., 2019; Qian et al., 2019; Qiu et al., 2018; Siravegna et al., 2017). At this time, however, based on an extensive joint review on ctDNA by the American Society of Clinical Oncology and the College of American Pathologist, there are still many questions regarding the clinical validity and clinical utility of ctDNA assays in cancer screening, early‐stage disease, and treatment monitoring (Merker et al., 2018). Further research, development of tools utilizing ctDNA, and clinical practice guidance are warranted. The liquid biopsy field will require further investigation with particular emphasis on clinical utility, not only clinical validation. This means demonstrating that liquid biopsy assay results will affect patient care in specific ways (e.g., changes in surgical approach and/or use of neoadjuvant therapies, changes in drug treatments) and that such changes will improve the morbidity and mortality of pancreatic cancer that we hope will occur in the near future. In particular, prospective clinical trials will be required that show that ctDNA may predict which patients are more likely to respond to chemotherapy and/or immune therapy and/or radiation therapy, or whether specific genetic aberrations identified in blood products (CTCs, ctDNA, EVs) predict response to particular therapies. For example, an exosomes test from Exosome Diagnostics is being tested as a companion diagnostic for Intezyne's phase 1/2 clinical trials of IT‐139, a novel cancer resistance pathway (CRP) inhibitor for the treatment of pancreatic, gastric, and other cancers in combination with existing anticancer therapies. Equally exciting it that the potential use of targeted EVs as a systemic treatment. A phase I trial that studies the best dose and side effects of mesenchymal stromal cells‐derived exosomes with KRAS G12D siRNA (iExosomes) has been approved by the U.S. National Cancer Institute (https://clinicaltrials.gov/ct2/show/NCT03608631) but not yet started. It will be used for the treatment of participants with metastatic pancreatic cancer with KRAS G12D mutation, hoping that iExosomes may prove a better treatment for this dismal disease. In summary, while liquid biopsy is promising, it still remains a burgeoning field. However, there are many positive signs that it will have a strong impact on the diagnosis, monitoring, and treatment of pancreatic cancer.

conflict of interest cc by

The authors declare no conflict of interest.

author contributions cc by

J‐SL and SSJ wrote the manuscript and created the figure and graphical abstract. SSP, YKL, and JAN edited the manuscript. All authors reviewed and approved the final version.