Pancreatic cancer is associated with a high mortality rate and represents the 7^th most frequent cause of cancer death, with approximately 265000 deaths and an incidence of 280000 per year worldwide in 2008[11,12]. Europe and Northern America have the highest incidence of pancreatic cancer, with slightly more males being affected.

Pancreatic cancer is usually diagnosed at an advanced stage due to a lack of symptoms in the early stages so that resection of the advanced tumor is often not possible. The overall 1-year survival rate for pancreatic cancer is 26%, and the 5-year survival rate is approximately 6% for advanced cancer and 22% for early stages when surgical removal of the tumor is still possible[13].

Therefore, new therapeutic approaches combining neoadjuvant chemotherapy and radiotherapy to significantly reduce the tumor size are promising to allow the option of surgical removal in selected patients[14].

Classical malignant pancreatic tumors show heterogeneous glandular and duct-like, grading-dependent structures, mostly infiltrating the pancreatic parenchyma and exhibiting a partially prominent desmoplastic stroma. The widely accepted and routinely used grading system of pancreatic cancer is based on (1) glandular differentiation; (2) mucin production; (3) mitosis (per 10 microscopic high power fields), and (4) nuclear features (such as nuclear polymorphism, size or arrangement of the nucleus)[15,16]. So far, no definitive and routinely used immunohistochemical markers exist, although many biological markers in pancreatic ductal adenocarcinoma were tested as possible diagnostic and prognostic tools. However, the main limitations arise from the small number of patients studied and in the heterogeneity of the applied methods[17]. Further approaches for prognostic grading focused on different morphological patterns similar to Gleason’s scoring system[18], included epithelial-mesenchymal characteristics, such as vimentin expression and tumor budding[19,20], or evaluated several gene expression signatures including downregulation of ASPM (abnormal spindle-like microcephaly associated) which could be detected by immunohistochemistry[21].

Detailed morphological analysis revealed prognostic subtypes of pancreatic cancer, with a group associated with better survival (colloid and medullary) and a group with a worse outcome (adenosquamous or undifferentiated)[22] (as described in Table 1).

It is important not only to give the diagnosis “pancreatic cancer”, but to discriminate between tumor entities for patient communication, and to evaluate a possible family history of cancer in genetic counselling (as in cases of medullary carcinoma[23]) as well as to establish tumor-specific therapy modalities, since it is possible to link tumor sub-entities to specific genetic lesions (Table 1).

Pancreatic intraepithelial neoplasia (PanIN), intraductal papillary mucinous neoplasms (IPMN) and mucinous cystic neoplasms (MCN) are considered precursor lesions[24] (Figure 1). As IPMN and MCN are seen pre-therapeutically in radiological investigations, therapeutic strategies/algorithms were established to weigh up the extent of surgical resection and patients’ quality of life.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Precursor lesions of pancreatic cancer with their typical morphologic and genetic as well epigenetic characteristics. PanIN: Pancreatic intraepithelial lesions; IPMN: Intraductal papillary mucinous neoplasms; MCN: Mucinous cystic neoplasms; RELN: Reelin.

Further molecular analyses of different precursor lesions and their morphological variants revealed a stepwise model of carcinogenesis and possible epigenetic associations (Figure 1). Interestingly, a huge number of epigenetically regulated genes have been detected by global gene expression profiles in comparison with classical genetic alterations showing an association with cytological and architectural atypia of these precursor lesions. The challenge in the future is to analyze the complex mechanistic crosstalk of these genetic and epigenetic regulatory mechanisms[25] during pancreatic carcinogenesis for new drug development and administration, as shown by Breitkreutz et al[26] using network statistics.

Intensive DNA analysis using genome- and epigenome-wide screening methods during the last few years have elucidated some major deregulated gate “drivers”, “passengers” and “keepers” in pancreatic cancer (Table 2)[27-30]. Overall, more suppressor genes than oncogenes are involved in pancreatic cancer.

As recently described in depth by Hong et al[31], the most frequently mutated oncogene (> 95%) in pancreatic cancer is K-RAS (Kirsten rat sarcoma viral oncogene homolog), leading to constitutive downstream signaling of proliferation, cellular survival, motility and remodeling. On the other hand, the major deregulated suppressor genes in pancreatic cancer are CDKN2A/p16 (cyclin-dependent kinase inhibitor 2A), TP53 (tumor protein p53) and SMAD4/DPC4 (SMAD family member 4) which are inactivated by 2 different, often independent, mechanisms. Whereas CDKN2A/p16 and TP53 are mainly involved in cell cycle checkpoint control and arrest, SMAD4/DPC4 plays an important role in signal transduction of the transforming growth factor (TGF)-β pathway, and furthermore in cellular proliferation. Finally, when looking at epigenetically affected genes, it is obvious that the classical and most frequent pancreatic cancer genes are only partially epigenetically regulated (Table 2). Interestingly, such cases with epigenetically silenced MLH1 (mutL homolog 1) genes are associated with the distinctive medullary phenotype of pancreatic cancer[32-34].

As later described in detail (Figure 1, Tables 3 and 4), epigenetics via DNA methylation, histone acetylation or interacting regulative microRNAs (miRNAs) could essentially be linked to different morphological and genetic changes during pancreatic carcinogenesis. Extensive investigations are been carried out on epigenetic changes in pancreatic cancer precursor lesions, indicating that heterogeneous, non-linked pathways of carcinogenesis are regulated by epigenetics as summarized in Figure 1 and Table 3. Detailed analysis of the function of these epigenetically deregulated genes revealed that all hallmarks of cancer[35,36] such as self-sufficiency in growth signals (e.g., SFRP1/SARP2; secreted frizzled-related protein 1), insensitivity to anti-growth signals [e.g., CDKN2A/p16 or RPRM (reprimo)], tissue invasion and metastasis [e.g., SPARC/ON (secreted protein, acidic, cysteine-rich (osteonectin)], limitless replicative potential [e.g., LHX1 (LIM homeobox 1)], sustained angiogenesis [e.g., CLDN5 (Claudin-5)] or evading apoptosis [e.g., RPRM or CADM1/TSLC1 (cell adhesion molecule 1/tumor suppressor in lung cancer 1)] are involved in pancreatic cancer and affected by epigenetic (de)regulation. This supports our knowledge of the pleiotropic effects of systemic epigenetic mechanisms. The degree of cytological and architectural atypia correlated with the amount of methylated genes supporting the hypothesized multiple step model of pancreatic cancer even in the early disease stages[31,37].

Additionally, hypomethylation is also recognized in pancreatic cancer, leading to genomic instability by overexpression of genes and proteins in contrast to hypermethylation by silencing genes and subsequent protein expression. Serial genes like SERPINB5 (serpin peptidase inhibitor, clade B, member 5), CLDN4, stratifin, lipocalin-2, trefoil factor2, S100P (S100 calcium-binding protein P), mesothelin or prostate stem cell antigen are hypomethylated which causes uncontrolled or “dys”-regulated cell cycle progression, proliferation, differentiation or adhesion[31].

The identification of DNA methylation, histone modification and the action of miRNAs has profoundly increased the knowledge about the regulation of gene activity. Epigenetics studies the stable and inheritable patterns of altered gene expression independent of the primary DNA sequence[38], and has shown that dynamic traits of chromatin, reversible covalent modification of DNA, and post-transcriptional regulation centrally impact on gene expression and phenotypic characteristics[8,39]. With increasing evidence that tumorigenesis-associated cellular changes are caused by epigenetic alterations, the field of cancer research has evolved to incorporate oncogenic mechanisms beyond DNA mutations. Epigenetic mechanisms (see Table 5 for an overview about the major epigenetic mechanisms) are generally reversible. Together with the fact that epigenetic alterations may be even more prevalent than genetic aberrations, this is highly attractive in the conceptual approach of selecting and exploiting potential molecular targets for novel cancer therapeutics[8,40].

The methylation of DNA and subsequent silencing of a gene is catalyzed by DNA methyltransferases (DNMTs) which add a methyl group to the 5’ carbon of the cytosine pyrimidine ring. This occurs preferably in regions containing cytosine-guanine dinucleotides (CpGs); these CpG islands are preferentially located in regions corresponding to regulatory regions of many genes[41]. While DNMT1 is responsible for maintenance of parental DNA methylation patterns following replication, de novo DNA methylation is catalyzed by DNMT3A and DNMT3B enzymes[42]. The identification of DNA demethylases which remove the methyl group and reverse the action of DNMTs still warrants further research. DNA methylation was the first type of epigenetic alteration identified as responsible for inactivation of a tumor suppressor gene[43], and it is suggested that 100-400 hypermethylated CpG islands may exist in a given tumor[44].

Compared with DNA-based epigenetics, alterations in DNA-associated histones offer a greater variety of covalent epigenetic modifications, including phosphorylation, methylation, acetylation, ubiquitination and sumoylation, all with different degrees of modification (e.g., mono-, di-, and trimethylation)[45,46]. The nucleosome as the core building block of eukaryotic chromatin comprises histone octamers (dimers of H2A, H2B, H3 and H4, i.e., the nucleosome core particle, NCP) and about 146 base pairs of DNA[47]. Modifications to histones determine the packing of DNA into either tight and transcriptionally silent heterochromatin or transcriptionally active and open-structured euchromatin. These modifications affecting chromatin mobility and stability have been termed “marks” whose type, position, and combinations determine whether a given gene is expressed or silenced, i.e., the histone code[46,48]. This hypothesis postulates that local histone modifications determine the configuration of chromatin, alter the binding affinities of non-histone proteins, and subsequently influence the chromatin structure and accessibility of DNA for transcription[39,49]. Histone acetylation is thought to activate gene transcription and is catalyzed by histone acetyltransferases (HATs) which transfer an acetyl group from acetyl coenzyme A to the ε-amino group of lysine such as CREBBP (cAMP response element-binding protein binding protein), p300 and p300-CBP-associated factor (P/CAF). The reverse reaction is mediated by histone deacetylases (HDACs) which comprise a group of 18 isoenzymes identified to date. HDACs are classified into four groups (I-IV) based on their homology to yeast HDACs as summarized in Table 5[50,51].

The polycomb group proteins (PcG) repress gene activity by trimethylation of H3-K27 (histone 3 lysine 27) while TrxG proteins (Trithorax group) activate gene expression via H3-K4 histone trimethylation. The PcGs have 2 functions: polycomb repressive complex 1 (PRC1) maintains the silenced (H3-K27-me3, trimethylated) chromatin state and consists of CBX-2, 4, or 8 (chromobox homologue 2/4/8), PHC-1, 2, or 3 (polyhomeotic homologue 1/2/3), BMI1 (B-cell-specific Moloney murine leukemia virus integration site 1), and RING1A/B or RNF2 (RING finger domain protein). PRC2 initiates the repressive state by de novo trimethylation of H3-K27 and consists of EZH2 (enhancer of zeste homologue 2), SUZ12 (suppressor of zeste 12) and EED (embryonic ectoderm development)[39]. Together with other chromatin-modifying enzymes including DNMTs and HDACs, the initially constituted suppression via H3K27-3me by PRC2 is maintained by PRC1 and allows fine-tuned, context-dependent regulation of gene silencing.

miRNAs are short (18-25 nucleotides), phylogenetically conserved single-stranded RNA molecules without protein-coding functions involved in the silencing of messenger RNAs (mRNAs)[52,53]. This post-transcriptional repression is accomplished by (partial) base-pairing with the respective mRNA causing either (1) inhibition of translation initiation; (2) inhibition of translation elongation; or (3) mRNA decay initiated by deadenylation of the mRNA following recruitment of a deadenylase complex[54,55]. The functions added a new layer of regulatory mechanisms affecting virtually all cellular functions[56,57]. The interaction between miRNAs and their target mRNAs ultimately leads to reduced levels of the regulated mRNA/protein. The current release of miRBase (v20[58]) lists 2578 mature miRNA sequences and it is estimated that over 60% of human mRNAs are direct miRNA targets[59]. miRNAs can function as either suppressors or oncogenes and their regulatory importance in human tumorigenesis has been demonstrated for various cancer types[60,61]. While miRNAs are per se categorized as an epigenetic mechanism, their cancer-related expression itself may be subject to epigenetic regulation by the above-mentioned mechanisms of chromatin modulation[62,63].

In the following paragraphs, we focus on their role in tumorigenesis in pancreatic cancer including their potential therapeutic exploitation by “epidrugs”. For a general overview on cancer-related epigenetic mechanisms, we kindly refer the reader to recent comprehensive reviews: DNA methylation[64-66], histone (de)acetylation[67-70], histone methylation[71-73], and miRNAs[74-76].

DNA methylation in pancreatic cancer: As discussed in McCleary-Wheeler et al[8] DNA methylation may occur early during tumorigenesis of pancreatic cancer as some epigenetic alterations are already observed in the earliest lesion, i.e., PanIN-1A, and their importance may increase during further tumor progression[30]. In line with these findings, mucin 4 (MUC4) gene expression is increased from normal to precancerous lesions to pancreatic cancer associated with an increasing frequency of MUC4 promoter hypomethylation[77]. Furthermore, not only DNA methylation but also hypomethylation and thus aberrantly high expression of a gene may represent an epigenetic feature in pancreatic carcinoma[78,79]. Table 3 summarizes the currently available literature on (deregulated) DNA methylation in pancreatic cancer.

Aberrant activation of developmental signalling pathways such as Hedgehog represents a frequently observed trait in human cancers[9]. He et al[80] have shown that the Hedgehog transcription factor Gli1 (GLI family zinc finger 1) targets epigenetic modifiers in pancreatic cancer, namely DNMT1 and DNMT3a. After showing a concomitantly higher expression of Gli1 and the DNMTs in pancreatic tumor samples, the authors proved by pharmacological inhibition using cyclopamine, Gli1 overexpression and siRNA (small interfering RNA)-based Gli1 knockdown, that the DNMT proteins are positive targets of this oncogenic pathway, suggesting a cross-talk between aberrantly activated embryogenesis pathways and activation of oncogenic epigenetic mechanisms in pancreatic cancer.

Related to the important role of epithelial-mesenchymal transition (EMT) in pancreatic tumor progression, FOXA1 and FOXA2 (forkhead box A1/2) transcriptions factor were identified as effective antagonists of EMT in pancreatic cancer due to their ability to positively regulate E-cadherin expression: in a study by Song et al[81], FOXA1/2 expression was lost during malignant progression and their promoter was extensively hypermethylated in a pancreatic cancer cell line. As the demethylation-mediated reactivation of E-cadherin required concomitant FOXA2 expression, the authors concluded that suppression of FOXA1/2 is necessary and sufficient for EMT in the progression of pancreatic cancer[81].

From the data summarized in Table 3 and the mentioned examples, it is clear that numerous genes and signaling targets are epigenetically regulated by means of DNA methylation in pancreatic cancer. Importantly, not only does late-stage, malignant pancreatic cancer display aberrant DNA methylation, but also earlier, pre-malignant lesions. It is important to note that DNA methylation may act in cooperation with other epigenetic mechanisms (histone modification) to achieve stable silencing of, for example, individual tumor suppressor genes. In a series of studies, Yamada et al[82-84] have shown that different mucin variants are regulated by different and complementary epigenetic mechanisms: MUC1, MUC2 and MUC5AC by DNA methylation and H3-K9 histone methylation. This fact has to be considered in therapeutic approaches which aim at reversing deregulated DNA methylation patterns, for example.

Histone-based epigenetics in pancreatic cancer: Acetylation and methylation of histones represent the 2 epigenetic mechanisms based on histone modifications for which currently data exist in the context of pancreatic tumorigenesis. Several studies have demonstrated the general relevance of HDAC enzymes in pancreatic cancer by showing that (1) HDAC2 is increased in pancreatic ductal adenocarcinoma, especially in poorly differentiated tumors[85]; and (2) the expression of HDAC7 is significantly increased in pancreatic adenocarcinoma samples; HDAC7 expression furthermore allows for discrimination between pancreatic adenocarcinoma from other pancreatic tumors (serous cystadenoma, IMPN)[86]. The potential of therapeutic targeting of HDACs in pancreatic cancers has been reviewed by Schneider et al[87]. As discussed in this section, several studies have unequivocally demonstrated the relevance of histone-based epigenetic mechanisms including the (oncogenic) functions of PRC1/2 protein complexes to contribute to pancreatic tumorigenesis. These studies provided evidence of either altered/aberrant expression of these epigenetic regulators in pancreatic cancer samples or demonstrated the potential therapeutic exploitation of these mechanisms using cell-based in vitro studies.

An interesting cooperation between the ZEB1 (zinc finger E-box binding homeobox) transcription factor which is responsible for silencing of the E-cadherin gene (CDH1) and HDAC enzymes was investigated by Aghdassi et al[88]. In 25 surgical pancreatic cancer specimens, the authors found neither hypermethylation nor somatic mutations in the CDH1 gene, but complexes of ZEB1/HDAC attached to the CDH1 promoter. Knockdown of ZEB1 prevented this interaction resulting in histone acetylation and re-expression of E-cadherin. This study has provided additional insights into how EMT transcription factors cooperate with HDACs to silence E-cadherin, and thus promote EMT and tumor progression[89]. These data confirmed earlier results which demonstrated that downregulation of E-cadherin in metastatic pancreatic cancer cells is mediated by a repressor complex containing the EMT transcription factor Snail and the HDAC1 and HDAC2 enzymes[90].

The particular role of PRC proteins as epigenetic mechanisms has only recently become a focus in pancreatic cancer research. In several tumor types, overexpression of the EZH2 gene is associated with poor prognosis, advanced stage, invasion and metastasis - for an overview see Crea et al[91]. As reviewed in Grzenda et al[39], EZH2 overexpression promoted survival, angiogenesis, migration, proliferation and repression of E-cadherin. For pancreatic cancer, several studies demonstrated an involvement of components of PRC1/2 in malignant progression: Martínez-Romero et al[92] found the expression of the PRC1 proteins Bmi1 and Ring1b to be upregulated in PanIN lesions (Bmi1) and pancreatic adenocarcinoma (both components). These finding were confirmed by Song et al[93] showing that Bmi1 is correlated with lymph node metastasis and poor survival rates; furthermore, stable knockdown of Bmi1 reduced the levels of cyclin D1, CDK-2/4 (cyclin-dependent kinase), Bcl-2 (B-cell CLL/lymphoma 2) and phosphor-Akt while the expression of p21 and Bax was increased and associated with higher susceptibility towards apoptosis induction. Moreover, Wellner et al[94] showed that Bmi1 and other “stemness” factors were negative targets of the epithelial differentiation-linked miR-200c, -203 and -183 suggesting a possible mechanism for Bmi1 deregulation in pancreatic cancer. Another PRC1 member was investigated by Karamitopoulou et al[95]: CBX7 was analyzed in 210 ductal pancreatic adenocarcinomas and its expression was found to progressively decrease from normal pancreatic tissue, PanINs and invasive ductal adenocarcinoma-associated with increasing malignancy and a trend to shorter overall survival.

Using immunohistochemistry, Ougolkov et al[96] could show nuclear overexpression of EZH2 in pancreatic cancer cell lines and in 68% (71/104) of pancreatic adenocarcinomas and that its nuclear accumulation was more prevalent in poorly differentiated pancreatic adenocarcinomas (91%, 31/34). Using RNA interference, genetic depletion of EZH2 sensitized pancreatic cancer cells to chemotherapy and caused re-expression of p27 [cyclin-dependent kinase inhibitor 1B (p27, Kip1)] and decreased proliferation. Similar results were obtained by Toll et al[97] who demonstrated an inverse relationship between expression of EZH2 and E-cadherin in 54 pancreatic adenocarcinomas, and significantly longer survival in gemcitabine-treated patients with low expression of EZH2. Interestingly, inactivation of RUNX3 (runt-related transcription factor 3), a component of TGF-β signalling, is mediated by at least 2 epigenetic mechanisms, both EZH2[98] and DNA hypermethylation[99], highlighting their cooperation to convey a malignant phenotype in pancreatic cancer. Recently, miR-218 was demonstrated to be negatively regulated by EZH2 in pancreatic ductal adenocarcinoma[100]. MiR-218 was significantly reduced in primary tumor samples compared with normal adjacent tissue, and its silencing was mediated by EZH2 which binds to the miR-218 promoter, promotes formation of heterochromatin, and recruits DNMT-1, -3A and -3B. MiR-218 expression reduces proliferation in vitro and tumor formation as well as metastasis in nude mice[100]. The inverse regulatory relationship has also been recently observed between EZH2 and pre-miR101. Overexpression of pre-miR-101 reduced the binding of EZH2 to the promoter of the E-cadherin gene (CDH1) and increased the levels of E-cadherin[101].

Epigenetics in pancreatic cancer stem cells: An interesting and possibly therapeutically relevant aspect is the role of EZH2 in maintenance of stemness characteristics in cancer, particular its role in maintaining the self-renewal capabilities of cancer stem cells (CSC)[102-108]. This subpopulation of cancer cells has been characterized in pancreatic cancer by surface markers CD44, CD24, CD133, ESA (EpCAM, epithelial cell adhesion molecule)[109,110] and is thought to represent the population of cancer cells responsible for tumor maintenance, tumorigenicity, metastasis, and resistance to conventional chemotherapeutic drugs, as well as recurrence[109,111]. Similar to studies with CSC derived from hepatocellular carcinoma and acute myeloid leukemia[104,112], recent studies have demonstrated the therapeutic potential of epidrugs to directly target this tumorigenic subpopulation of pancreatic cancer cells: Avan et al[102] used an inhibitor of the EZH2 methyltransferase (DZNep, deazaneplanocin-A) and showed that treatment with DZNep reduced spheroid formation of pancreatic cancer cells and decreased the CD133⁺ subpopulation. Furthermore, the combination of DZNep and gemcitabine was shown to be highly synergistic and was accompanied by a reduced percentage of G2/M cells, reduced migration, increased E-cadherin expression and increased apoptosis[102]. The level of EZH2 has furthermore been suggested as an assay to effectively measure changes in the CSC subpopulation: using pancreatic and breast cancer cell lines, knockdown of EZH2 by RNA interference decreased the CSC subpopulation, confirming its role in CSC maintenance, and genes affected by EZH2 knockdown were inversely correlated with their expression in enriched CSC subpopulations[108]. The Hedgehog pathway has also been implicated in the maintenance of CSCs in various models[9,113]; interestingly a combination of Hedgehog inhibition (SANT-1) and SAHA (a pan-HDAC inhibitor; suberanilohydroxamic acid, Vorinostat) synergistically suppressed proliferation and colony formation in gemcitabine-resistant pancreatic adenocarcinoma cell lines by increased Bax expression, activation of caspase-3/7, increased p21 and p27 and reduced cyclin D1 expression. This study suggests that combined inhibition of stem cell-associated pathways (Hedgehog) and epigenetic drugs could be efficient in targeting the CSC subpopulation in pancreatic cancer[114]. A study by Nalls et al[115] could demonstrate that demethylating agents (5-aza-dC, 5-aza-2’-deoxycytidine) and the HDAC inhibitor SAHA restored the expression levels of miR-34a, which is reduced in pancreatic CSCs. These inhibitors caused a reduction in the EMT-related ZEB1, Snail, and Slug transcription factors, increased epithelial marker expression (E-cadherin) and, most importantly, reduced the number of viable pancreatic CSC, accompanied by reduced migration, colony formation and invasion of these cells.

Based on the above-mentioned functions and properties of CSC which is critical for tumor initiation, metastasis, progression and therapeutic resistance, these findings are of central importance and warrant further investigation to hopefully develop (epigenetics-based) therapeutic regimens specifically targeting this tumorigenic subpopulation in pancreatic cancer.

miRNA-based epigenetics in pancreatic cancer: Some reporters[116-119] and Park et al[120] have reviewed the available publications on differential miRNA expression in pancreatic cancer vs normal tissue culminating in a list of 64, partly overlapping individual miRNAs which were found to be deregulated in pancreatic cancer. Of these miRNAs, overexpression of miR-21, -155, -196a-2, -203, -210 and -222 was furthermore associated with poor outcome[120].

Table 4 provides an update (based on Park et al[120] 2011) of the currently available literature on the specific role of individual miRNAs in pancreatic cancer. All of these studies investigated the cellular/molecular mechanisms of the oncogenic or tumor-suppressive action of miRNAs, mainly by forced overexpression or knockdown of the respective miRNAs.

An example of how epigenetic mechanisms are employed to regulate the expression of tumor-suppressive miRNAs is shown in the study of Zhang et al[121]. From 12 miRNAs differentially expressed in pancreatic cancers vs adjacent normal tissue, miR-132 was downregulated in 16/20 pancreatic carcinomas accompanied by methylation of its promoter, as shown both in cell lines and tumor tissue. Sp-1 expression was correlated with miR-132 expression, and its binding affinity to the miR-132 promoter was significantly lower in pancreatic tumors relative to non-tumor samples.

As recently discussed[120,122], epigenetic features and especially miRNAs could also serve as biomarkers to allow specific and sensitive diagnosis of pancreatic cancer - an important approach as most patients with this disease remain without symptoms until the lesion has progressed to an advanced or metastatic stage. In this context, the use of miR-155 which showed upregulation in most IPMNs (83% of cases) has been analyzed in pancreatic juice by Habbe et al[123]. The authors confirmed upregulation of the miR-155 transcript in 60% (6/10) of IPMN-associated pancreatic juice samples but in none of the 5 control cases. Wang et al[124] profiled 4 miRNAs (miR-21, -210, -155, and -196a) in heparin-treated blood samples and found a sensitivity of 64% and a specificity of 89% to distinguish pancreatic cancer patients from healthy controls using this panel of miRNAs, thus proving the feasibility of plasma-based miRNA profiling as a potential biomarker for pancreatic cancer. Furthermore, Kawaguchi et al[125] investigated the utility of plasma miR-221 as a biomarker for cancer detection and monitoring tumor dynamics in 47 consecutive pancreatic cancer patients: similar to cancer tissue, plasma miR-221 levels were significantly higher in pancreatic cancer patients and correlated with distant metastasis and non-resectable status. Also, miR-21 serum levels were shown to be associated with overall survival of pancreatic cancer patients, and, in combination with 6 other miRNAs, allowed for correct classification of clinically suspected pancreatic cancer with a rate of 84%[126]. Similar results were obtained for miR-18a in plasma samples of patients with pancreatic cancer: miR-18a levels were significantly higher in 36 cancer patients compared with 30 healthy volunteers[127]. Kong et al[128] investigated the utility of several miRNAs as serum markers: while miR-21 distinguished pancreatic ductal adenocarcinoma patients from chronic pancreatitis and controls, miR-196a could distinguish resectable (stages I and II) and unresectable pancreatic ductal adenocarcinoma (III and IV) as well as predict median survival time of pancreatic ductal adenocarcinoma patients (6.1 mo vs 12.0 mo for high vs low level miR-196a). Recently, miR-21 from pancreatic cyst fluid was investigated as a potential biomarker and could differentiate between benign, premalignant and malignant pancreatic cyst neoplasms[129].